Description:FORM 2

THE PATENTS ACT, 1970
(39 of 1970)
&
THE PATENT RULES, 2003

COMPLETE SPECIFICATION
(See Section 10 and Rule 13)

Title of invention:

METHOD AND SYSTEM FOR CONTEXT SPECIFIC LANGUAGE TRANSLATION

Applicant

Tata Consultancy Services Limited
A company Incorporated in India under the Companies Act, 1956
Having address:
Nirmal Building, 9th floor,
Nariman point, Mumbai 400021,
Maharashtra, India

Preamble to the description:

The following specification particularly describes the invention and the manner in which it is to be performed.
TECHNICAL FIELD
[001] The disclosure herein generally relates to document processing, and, more particularly, to a method and system for context specific language translation.

BACKGROUND
[002] In a world with diverse languages, images, infographics & currencies it is often difficult to generate insights out of the scattered information and data points across the most widely spoken languages of the world. As a part of sales & strategic enablement it is becomes difficult to collate, analyze, and curate data points and information across different languages to contextualized for decision making. Since many of the documents and responses are available in different languages as per the geographic specific importance, this hampers the agility with respect to time & requires huge investments from part of the Corporates to generate contextualized data for decision making. The tools/processes/steps/frameworks/models available for similar type of activities are limited to or could not perform the idiomatic, images, expressions, currency or voice translations all at the same time. They do literal translation which differs from actual meaning. Also, models available in the market for commercial purpose and in research state, have limited capabilities in terms of parallel processing of translation related activities.

SUMMARY
[003] Embodiments of the present disclosure present technological improvements as solutions to one or more of the above-mentioned technical problems recognized by the inventors in conventional systems. For example, in one embodiment, a processor implemented method is provided. In this method, a document is fetched as input, via one or more hardware processors. Further, one or more data types in the document are determined via the one or more hardware processors, wherein by determining the one or more data types in the document, data in the document is segregated as data that can be processed and data that cannot be processed. Further, the data that has been segregated as the data that can be processed is translated using a data processing model, via the one or more hardware processors. Translating the data using the data model includes performing the following steps till a measured confidence score at least matches a threshold of confidence score. In this process, initially a current language of the document and a target language are determined. Further, the data is segregated as one or more of a text component, an audio/video (av) component, an infographics component, and a currency component. Further, the text component is processed using a text translation technique, to convert the data in the text component from the current language to the target language. Further, the a/v component is processed using an a/v translation technique, to convert the data in the a/v component from the current language to the target language. Further, the infographics component is processed using an infographics translation technique, to convert the data in the infographics component from the current language to the target language. Further, the currency component is processed using a currency translation technique, to convert the data in the currency component from the current language to the target language.
[004] In another aspect, the method includes processing the text component using the text translation technique comprises translation of text, infographics, phases, Parts of Speech (PoS), and idioms, in the text component to the target language.
[005] In another aspect, wherein processing the infographics component using the infographics translation technique in the method includes the following steps. Initially, one or more infographics in the infographics component are scanned to determine a plurality of word contours (WC) and alphabet contours (AC). Further, the plurality of WCs and ACs are segmented and classified as different PoS, to generate a plurality of segments. Further, each of the plurality of segments is labelled and mapped to words present in each of a plurality of languages stored in a reference database. Further, word matching the one or more infographics is identified by comparing each of a plurality of combinations of the plurality of segments and the words with a plurality of reference data pointers. Further, identified words are translated to the target language, based on a determined context of the document fetched as input and an associated emotion of a text being translated, wherein the associated emotion is determined in terms of one or more cultural proximities of the current language.
[006] In another aspect, in this method, the reference database is generated using a corpus based sentiment analysis on data comprising semantics of a plurality of words from one or more languages, in a plurality of iterations, wherein in each of the plurality of iterations, associations between the plurality of words is established and updated, wherein a lexicon based dynamic crawling method is used to determine sentiment of words during the sentiment analysis.
[007] In another aspect, the method of processing the a/v component using the a/v translation technique includes the following steps. Initially, modulation of language in the a/v component along with frequency of associated waves is captured. Further, a plurality of word contours (WC) and alphabet contours (AC) in the a/v component are identified, based on the captured modulation of language. Further, one or more words matching the a/v component are identified by mapping the WCs and ACs to a plurality of pointers in a reference data. Further, the identified one or more words are translated to the target language, based on a determined context of the document fetched as input and an associated emotion of a text being translated.
[008] In another aspect, the method of processing the currency component using the currency translation technique includes the following steps. Initially, by processing the document, a currency type and a currency value are determined, wherein the currency type is determined based on a) a currency type specified adjacent to the currency value, or b) based on the determined current language of the document. Further, based on a received user input, one of a) a current date, b) a date of compilation of the document, or c) a selected date, is determined as baseline for currency conversion. Further, a currency conversion rate with respect to the determined baseline is determined, and then the currency conversion is performed based on the currency value and the determined currency conversion rate.
[009] In another aspect, the method of determining the current language by processing the document fetched as input, includes the following steps. Initially, a plurality of words from the document are compared with words in a plurality of languages in a reference database, wherein the plurality of words is selected from a portion of the document identified using an adaptive progress language technique. Further, a similarity percentage for each of the plurality of languages is determined based on extent of similarity of the plurality of words with each of the plurality of languages. Further, a language is determined from among the plurality of languages, having the similarity percentage exceeding a threshold of similarity, as the current language.
[010] In another aspect, a system is provided. The system includes one or more hardware processors, a communication interface, a memory storing a plurality of instructions. The plurality of instructions when executed, cause the one or more hardware processors to fetch a document as input. Further, one or more data types in the document are determined via the one or more hardware processors, wherein by determining the one or more data types in the document, data in the document is segregated as data that can be processed and data that cannot be processed. Further, the data that has been segregated as the data that can be processed is translated using a data processing model, via the one or more hardware processors. Translating the data using the data model includes performing the following steps till a measured confidence score at least matches a threshold of confidence score. In this process, initially a current language of the document and a target language are determined. Further, the data is segregated as one or more of a text component, an audio/video (av) component, an infographics component, and a currency component. Further, the text component is processed using a text translation technique, to convert the data in the text component from the current language to the target language. Further, the a/v component is processed using an a/v translation technique, to convert the data in the a/v component from the current language to the target language. Further, the infographics component is processed using an infographics translation technique, to convert the data in the infographics component from the current language to the target language. Further, the currency component is processed using a currency translation technique, to convert the data in the currency component from the current language to the target language.
[011] In yet another aspect, processing the text component using the text translation technique by the system includes translation of text, infographics, phases, Parts of Speech (PoS), and idioms, in the text component to the target language.
[012] In yet another aspect, wherein processing the infographics component using the infographics translation technique by the system includes the following steps. Initially, one or more infographics in the infographics component are scanned to determine a plurality of word contours (WC) and alphabet contours (AC). Further, the plurality of WCs and ACs are segmented and classified as different PoS, to generate a plurality of segments. Further, each of the plurality of segments is labelled and mapped to words present in each of a plurality of languages stored in a reference database. Further, word matching the one or more infographics is identified by comparing each of a plurality of combinations of the plurality of segments and the words with a plurality of reference data pointers. Further, identified words are translated to the target language, based on a determined context of the document fetched as input and an associated emotion of a text being translated, wherein the associated emotion is determined in terms of one or more cultural proximities of the current language.
[013] In yet another aspect, generating the reference database using a corpus based sentiment analysis on data comprising semantics of a plurality of words from one or more languages, by the system, includes performing in a plurality of iterations, wherein in each of the plurality of iterations, associations between the plurality of words is established and updated, wherein a lexicon based dynamic crawling method is used to determine sentiment of words during the sentiment analysis.
[014] In yet another aspect, processing the a/v component using the a/v translation technique by the system includes the following steps. Initially, modulation of language in the a/v component along with frequency of associated waves is captured. Further, a plurality of word contours (WC) and alphabet contours (AC) in the a/v component are identified, based on the captured modulation of language. Further, one or more words matching the a/v component are identified by mapping the WCs and ACs to a plurality of pointers in a reference data. Further, the identified one or more words are translated to the target language, based on a determined context of the document fetched as input and an associated emotion of a text being translated.
[015] In yet another embodiment, wherein processing the currency component using the currency translation technique, by the system, includes the following steps. Initially, by processing the document, a currency type and a currency value are determined, wherein the currency type is determined based on a) a currency type specified adjacent to the currency value, or b) based on the determined current language of the document. Further, based on a received user input, one of a) a current date, b) a date of compilation of the document, or c) a selected date, is determined as baseline for currency conversion. Further, a currency conversion rate with respect to the determined baseline is determined, and then the currency conversion is performed based on the currency value and the determined currency conversion rate.
[016] In yet another aspect, determining the current language by processing the document fetched as input, by the system, includes the following steps. Initially, a plurality of words from the document are compared with words in a plurality of languages in a reference database, wherein the plurality of words is selected from a portion of the document identified using an adaptive progress language technique. Further, a similarity percentage for each of the plurality of languages is determined based on extent of similarity of the plurality of words with each of the plurality of languages. Further, a language is determined from among the plurality of languages, having the similarity percentage exceeding a threshold of similarity, as the current language.
[017] In yet another aspect, a non-transitory computer readable medium is provided. The non-transitory computer readable medium includes a plurality of instructions, which when executed by one or more hardware processors, causes execution of the following steps. Initially, a document is fetched as input, via one or more hardware processors. Further, one or more data types in the document are determined via the one or more hardware processors, wherein by determining the one or more data types in the document, data in the document is segregated as data that can be processed and data that cannot be processed. Further, the data that has been segregated as the data that can be processed is translated using a data processing model, via the one or more hardware processors. Translating the data using the data model includes performing the following steps till a measured confidence score at least matches a threshold of confidence score. In this process, initially a current language of the document and a target language are determined. Further, the data is segregated as one or more of a text component, an audio/video (av) component, an infographics component, and a currency component. Further, the text component is processed using a text translation technique, to convert the data in the text component from the current language to the target language. Further, the a/v component is processed using an a/v translation technique, to convert the data in the a/v component from the current language to the target language. Further, the infographics component is processed using an infographics translation technique, to convert the data in the infographics component from the current language to the target language. Further, the currency component is processed using a currency translation technique, to convert the data in the currency component from the current language to the target language.
[018] In yet another aspect, the non-transitory computer readable medium configures the one or more hardware processors to process the text component using the text translation technique by performing translation of text, infographics, phases, Parts of Speech (PoS), and idioms, in the text component to the target language.
[019] In yet another aspect, the non-transitory computer readable medium configures the one or more hardware processors to perform processing of the infographics component using the infographics translation technique by performing the following steps. Initially, one or more infographics in the infographics component are scanned to determine a plurality of word contours (WC) and alphabet contours (AC). Further, the plurality of WCs and ACs are segmented and classified as different PoS, to generate a plurality of segments. Further, each of the plurality of segments is labelled and mapped to words present in each of a plurality of languages stored in a reference database. Further, word matching the one or more infographics is identified by comparing each of a plurality of combinations of the plurality of segments and the words with a plurality of reference data pointers. Further, identified words are translated to the target language, based on a determined context of the document fetched as input and an associated emotion of a text being translated, wherein the associated emotion is determined in terms of one or more cultural proximities of the current language.
[020] In yet another aspect, the non-transitory computer readable medium configures the one or more hardware processors to generate the reference database using a corpus based sentiment analysis on data comprising semantics of a plurality of words from one or more languages, in a plurality of iterations, wherein in each of the plurality of iterations, associations between the plurality of words is established and updated, wherein a lexicon based dynamic crawling method is used to determine sentiment of words during the sentiment analysis.
[021] In yet another aspect, the non-transitory computer readable medium configures the one or more hardware processors to process the a/v component using the a/v translation technique by performing the following steps. Initially, modulation of language in the a/v component along with frequency of associated waves is captured. Further, a plurality of word contours (WC) and alphabet contours (AC) in the a/v component are identified, based on the captured modulation of language. Further, one or more words matching the a/v component are identified by mapping the WCs and ACs to a plurality of pointers in a reference data. Further, the identified one or more words are translated to the target language, based on a determined context of the document fetched as input and an associated emotion of a text being translated.
[022] In yet another aspect, the non-transitory computer readable medium configures the one or more hardware processors to processing the currency component using the currency translation technique by performing the following steps. Initially, by processing the document, a currency type and a currency value are determined, wherein the currency type is determined based on a) a currency type specified adjacent to the currency value, or b) based on the determined current language of the document. Further, based on a received user input, one of a) a current date, b) a date of compilation of the document, or c) a selected date, is determined as baseline for currency conversion. Further, a currency conversion rate with respect to the determined baseline is determined, and then the currency conversion is performed based on the currency value and the determined currency conversion rate.
[023] In yet another aspect, the non-transitory computer readable medium configures the one or more hardware processors to perform determining the current language by processing the document fetched as input, by performing the following steps. Initially, a plurality of words from the document are compared with words in a plurality of languages in a reference database, wherein the plurality of words is selected from a portion of the document identified using an adaptive progress language technique. Further, a similarity percentage for each of the plurality of languages is determined based on extent of similarity of the plurality of words with each of the plurality of languages. Further, a language is determined from among the plurality of languages, having the similarity percentage exceeding a threshold of similarity, as the current language.
[024] It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the invention, as claimed.

BRIEF DESCRIPTION OF THE DRAWINGS
[025] The accompanying drawings, which are incorporated in and constitute a part of this disclosure, illustrate exemplary embodiments and, together with the description, serve to explain the disclosed principles:
[026] FIG. 1 illustrates an exemplary system for language translation, according to some embodiments of the present disclosure.
[027] FIG. 2 is a flow diagram depicting steps involved in the process of performing the language translation, by the system of FIG. 1, according to some embodiments of the present disclosure.
[028] FIG. 3 is a flow diagram depicting steps involved in the process of translating data using a data processing model, by the system of FIG. 1, in accordance with some embodiments of the present disclosure.
[029] FIG. 4 is a flow diagram depicting steps involved in the process of processing an infographics component using the infographics translation technique, by the system of FIG. 1, in accordance with some embodiments of the present disclosure.
[030] FIG. 5 is a flow diagram depicting steps involved in the process of processing an audio/video (a/v) component, using an a/v translation technique, by the system of FIG. 1, in accordance with some embodiments of the present disclosure.
[031] FIG. 6 is a flow diagram depicting steps involved in the process of processing a currency component, using a currency translation technique, by the system of FIG. 1, in accordance with some embodiments of the present disclosure.
[032] FIG. 7 is a flow diagram depicting steps involved in the process of determining a current language of input document, by the system of FIG. 1, in accordance with some embodiments of the present disclosure.

DETAILED DESCRIPTION OF EMBODIMENTS
[033] Exemplary embodiments are described with reference to the accompanying drawings. In the figures, the left-most digit(s) of a reference number identifies the figure in which the reference number first appears. Wherever convenient, the same reference numbers are used throughout the drawings to refer to the same or like parts. While examples and features of disclosed principles are described herein, modifications, adaptations, and other implementations are possible without departing from the scope of the disclosed embodiments.
[034] As a part of sales & strategic enablement it is becomes difficult to collate, analyze, and curate data points and information across different languages to be contextualized for decision making. Since many of the documents and responses are available in different languages as per the geographic specific importance. This hampers the agility with respect to time and requires huge investments from part of the corporates to generate contextualized data for decision making. The tools/processes/steps/frameworks/models available for similar type of activities are limited to or could not perform the idiomatic, images, expressions, currency or voice translations all at the same time. Also, models available in the market for commercial purpose and in research state, have limited capabilities in terms of parallel processing of translation related activities.
[035] In order to address these challenges, a method and system for context specific language translation is provided. In this approach, a document is fetched as input, via one or more hardware processors. Further, one or more data types in the document are determined via the one or more hardware processors, wherein by determining the one or more data types in the document, data in the document is segregated as data that can be processed and data that cannot be processed. Further, the data that has been segregated as the data that can be processed is translated using a data processing model, via the one or more hardware processors. Translating the data using the data model includes performing the following steps till a measured confidence score at least matches a threshold of confidence score. In this process, initially a current language of the document and a target language are determined. Further, the data is segregated as one or more of a text component, an audio/video (av) component, an infographics component, and a currency component. Further, the text component is processed using a text translation technique, to convert the data in the text component from the current language to the target language. Further, the a/v component is processed using an a/v translation technique, to convert the data in the a/v component from the current language to the target language. Further, the infographics component is processed using an infographics translation technique, to convert the data in the infographics component from the current language to the target language. Further, the currency component is processed using a currency translation technique, to convert the data in the currency component from the current language to the target language.
[036] Referring now to the drawings, and more particularly to FIG. 1 through FIG. 7, where similar reference characters denote corresponding features consistently throughout the figures, there are shown preferred embodiments and these embodiments are described in the context of the following exemplary system and/or method.
[037] FIG. 1 illustrates an exemplary system for language translation, according to some embodiments of the present disclosure. The system 100 includes or is otherwise in communication with hardware processors 102, at least one memory such as a memory 104, an I/O interface 112. The hardware processors 102, memory 104, and the Input /Output (I/O) interface 112 may be coupled by a system bus such as a system bus 108 or a similar mechanism. In an embodiment, the hardware processors 102 can be one or more hardware processors.
[038] The I/O interface 112 may include a variety of software and hardware interfaces, for example, a web interface, a graphical user interface, and the like. The I/O interface 112 may include a variety of software and hardware interfaces, for example, interfaces for peripheral device(s), such as a keyboard, a mouse, an external memory, a printer and the like. Further, the I/O interface 112 may enable the system 100 to communicate with other devices, such as web servers, and external databases.
[039] The I/O interface 112 can facilitate multiple communications within a wide variety of networks and protocol types, including wired networks, for example, local area network (LAN), cable, etc., and wireless networks, such as Wireless LAN (WLAN), cellular, or satellite. For the purpose, the I/O interface 112 may include one or more ports for connecting several computing systems with one another or to another server computer. The I/O interface 112 may include one or more ports for connecting several devices to one another or to another server.
[040] The one or more hardware processors 102 may be implemented as one or more microprocessors, microcomputers, microcontrollers, digital signal processors, central processing units, node machines, logic circuitries, and/or any devices that manipulate signals based on operational instructions. Among other capabilities, the one or more hardware processors 102 is configured to fetch and execute computer-readable instructions stored in the memory 104.
[041] The memory 104 may include any computer-readable medium known in the art including, for example, volatile memory, such as static random-access memory (SRAM) and dynamic random-access memory (DRAM), and/or non-volatile memory, such as read only memory (ROM), erasable programmable ROM, flash memories, hard disks, optical disks, and magnetic tapes. In an embodiment, the memory 104 includes a plurality of modules 106.
[042] The plurality of modules 106 include programs or coded instructions that supplement applications or functions performed by the system 100 for executing different steps involved in the process of switching between hardware accelerators for model training, being performed by the system 100. The plurality of modules 106, amongst other things, can include routines, programs, objects, components, and data structures, which performs particular tasks or implement particular abstract data types. The plurality of modules 106 may also be used as, signal processor(s), node machine(s), logic circuitries, and/or any other device or component that manipulates signals based on operational instructions. Further, the plurality of modules 106 can be used by hardware, by computer-readable instructions executed by the one or more hardware processors 102, or by a combination thereof. The plurality of modules 106 can include various sub-modules (not shown). The plurality of modules 106 may include computer-readable instructions that supplement applications or functions performed by the system 100 for the switching between hardware accelerators for model training.
[043] The data repository (or repository) 110 may include a plurality of abstracted piece of code for refinement and data that is processed, received, or generated as a result of the execution of the plurality of modules in the module(s) 106.
[044] Although the data repository 110 is shown internal to the system 100, it will be noted that, in alternate embodiments, the data repository 110 can also be implemented external to the system 100, where the data repository 110 may be stored within a database (repository 110) communicatively coupled to the system 100. The data contained within such external database may be periodically updated. For example, new data may be added into the database (not shown in FIG. 1) and/or existing data may be modified and/or non-useful data may be deleted from the database. In one example, the data may be stored in an external system, such as a Lightweight Directory Access Protocol (LDAP) directory and a Relational Database Management System (RDBMS). Functions of the components of the system 100 are now explained with reference to the steps in flow diagrams in FIG. 2 through FIG. 7.
[045] FIG. 2 is a flow diagram depicting steps involved in the process of performing the language translation, by the system of FIG. 1, according to some embodiments of the present disclosure.
[046] At step 202 of the method 200, a document is fetched as input, via the one or more hardware processors 102. The document may include data such as but not limited to text information from large contextual text, documents or reports or extractions from links, currency value, extraction of text from infographics, and non-verbal communication and signs. The input maybe a) Text (Includes Phrases, Idioms, Cultural Phrases, non-verbal Signs) b) Textual Images c) GIFs d) Documents (pdf, doc, ppt format) e) Voices. This means that the document may contain data of different data types. In order to process the document further, it is required to extract data from the document and identify different data types. This is done at step 204 of the method 200. In an embodiment, the system 100 may use appropriate text extraction techniques such as but not limited to optical character recognition (OCR) for extracting text from the document. Phrases or group of words or parts of speech also are detected using appropriate techniques, for example, using Hidden Markov Modelling (HMM) technique.
[047] Processing of the document and extraction of the data using the HMM technique may involve identification of parts of speech, which helps in segregation of groups of words or next sequence as required. Post segregation of the groups of words, the system 100 may associate different types of probabilities, for example, i) Transition Probabilities (TP - which determines the probability of the next word) and ii) Emission Probabilities (EP - which determines the associated word if any), with the extracted data. Example of associating different groups of words with different probabilities is given in Table. 1.
Parts of Speech Noun Verb Preposition Pronoun Adverb Adjective Conjunction Interjection
Tag n v pre pro adv adj c i
TP TP1 TP2 TP3 TP4 TP5 TP6 TP7 TP8
EP EP1 EP2 EP3 EP4 EP5 EP6 EP7 EP8
Output TP1-n-EP1 TP2-v-EP2 ____ ____ ____ ____ ____ ____

Table. 1
[048] Similarly, the document may contain data in the form of, for example, a combination of alphabets showcasing a sign. In this case if there is no meaning to the word that is being translated, the system 100 may keep this data as is, i.e. without any change. Similarly, if there is mix of text and signs which are not getting translated to any meaning, it is converted to image by using any suitable technique such as an open-source algorithm, and maybe considered as non-verbal communication signs, which maybe then translated.
[049] Similarly, the document may contain some images with specific meaning. For example, a traffic cop giving certain hand signals for regulating traffic flow. Meaning of such images maybe interpreted based on position of different body parts (for example, hands, legs, head/face), which maybe then compared with a corresponding reference data to interpret direct meaning, if any. For the purpose of interpreting meaning of such images, the system 100 may use data from a reference database as given in Table. 2.
Body Parts Head or Face Eyes Nose Mouth Tongue Hand or Arms Fingers Body Back Legs or Feet Toes Others
Symbols HF EY NO MO TO HA FI BO BA LF TO OT

Table. 2
[050] Processing of the image based on the reference data in the table involves classifying different combinations of the images to depict non-verbal communications. As given in Table. 2, there are 12 types of body parts, which could be moved to produce different non-verbal communications. There could be 12! ways of expressions with the help of the considered body parts. Also, different types of categorizations (9, in this example) – time, space, physical, body movements, characteristic, paralanguage, artefacts, environment, color, and touch. Hence, in total there could be 12! *9! ways of expressions when it comes to expression of the body part, in an image processing approach used by the system 100. While processing the images, each image within the text or derived from text representing a non-verbal communication sign is divided into pixels and each pixel and is given a co-ordinate R(I,J), where I represents the distance from the X-Axis in a 2D Place and J represents the distance from the Y-Axis in a 2D plane. The combination of 12! *9! Along with R(I, J) pointers are mapped in the background for each language in order to classify the different non-verbal communication signs. This helps in constructing the set of point meshes or point clouds. For any language, a specific combination of the pixels for a non-verbal communication sign is generated, which is mapped with non-verbal communication signs of the other languages. Different local descriptors or vectors in an input language cloud mesh mapping help in translation by forming a cloud of local descriptor points, which in turn results in developing communication sign for the output language from the meaning of that cloud in input language. For example, if in one language “Namaste” which is a sign of closed hands gets translated to Japanese, the folded hands image will be mapped to cloud points with local descriptors. The meaning of this in Hindi language is greeting and thereby it will get converted to sign of ‘Bended Body’ in Japanese language, so that contextualization also is achieved. This example is depicted in Table. 3.
Converted Sign Classifier Instance Context Intermediate Output Corresponding
Emotion 1 Namaste (Hindi) At the beginning of a conversation Greetings (English) Welcome Symbol or only sign-welcome
Emotion 2 Namaste (Hindi) At the end of a conversation Thank You (English) Shake Hands Symbol or only sign-Thank You
…. …. …. ….
Emotion N …. …. ….

Table. 3
[051] In absence of any direct mapping, the meaning of that communication in the input language is being translated to literal meaning of the communication in the target language and a prefix called ‘Sign’ is attached with the meaning. The images extracted from text or direct images within text are being compared with the same information in the stored database. This is done by highlighting the interested pointers within a 2-Dimensional (2D) plane as per the language or the culture.
[052] By determining the one or more data types in the document, data in the document is segregated as data that can be processed and data that cannot be processed. In the context of the embodiments disclosed herein, the data that has been segregated as the data that can be processed are data for which associated data type could be recognized by the system 100, and for which at least one data processing technique is configured with the system 100. Similarly, the data that has been segregated as the data that cannot be processed are data for which associated data type could not be recognized by the system 100, and for which no data processing technique is configured with the system 100.
[053] In an embodiment, the system 100 may store in an associated database, information on all the types of data that can be processed by the system 100 and all the types of data that cannot be processed by the system 100, which the system 100 may use as a reference for segregating the data as the data that can be processed and the data that cannot be processed. An example representation of the reference data is given in Table. 4.
Inputs
Any format that could be processed further Any format that could not be processed further
Input 1 ---- ----
Input 2 Yes ----
Input 3 ---- ----
Input 4 ---- Yes
Input … ---- ----
Input N ---- ----

Table. 4

[054] Further, at step 206 of the method 200, the data that has been segregated as the data that can be processed is translated using a data processing model, via the one or more hardware processors 102. Translating the data using the data model includes performing steps 302 through 312 in method 300 in FIG. 3, till a measured confidence score at least matches a threshold of confidence score. The confidence score is measured using standard confidence interval measurement statistical technique.
[055] At step 302 of the method 300, a current language of the document and a target language are determined. In an embodiment, the current language is automatically determined by the system 100, by processing the document. Steps in determining the current language of the document by the system 100 are depicted in method 700 in FIG. 7, and are explained hereafter. At step 702 of the method 700, a plurality of words from the document are compared with words in a plurality of languages in a reference database, wherein the plurality of words is selected from a portion of the document identified using an adaptive progress language technique. The adaptive progress language technique involving determining by the system 100 that data required for the comparison of the document with different languages can be done based on data from a certain percentage of the document i.e. 20%, 30% and so on.
[056] In the adaptive progress language technique, the system 100 compares a group of word(s) [two or three or four or five as per the availability of the words] in the document with the stored database of language and thereby determining a single language to check back with the original text. Here use of space is being used to segregate the words and use of ‘-‘ being used to understand the hyphenated words and for special signs as prefix or superscript or under script of the words, languages are further filtered. An example of the selection of words in the adaptive progress language technique is given in Table. 5.

Inputs Text Group of words Categories of signs Certain percentage
Input 1 Group of Words 5 Groups Superscript Presence 1.7%
Input 2 Group of Words 6 Groups Superscript Presence 2.5%
Input 3 Group of Words 8 Groups Superscript Presence 1.7%
Input 4 Group of Words 2 Groups Superscript Presence 90%
Input … ---- ----
Input N Group of Words 5 Groups Superscript Presence 2.5%

Table. 5
[057] Further, at step 704 of the method 700, a similarity percentage for each of the plurality of languages is determined based on extent of similarity of the plurality of words with each of the plurality of languages. Further, at step 706 of the method 700, a language having the similarity percentage exceeding a threshold of similarity from among the plurality of languages, is determined as the current language. In an embodiment, if for more than one language, the similarity percentage exceeds the threshold of similarity, the language having highest value of the similarity percentage is determined as the current language. In another embodiment, the current language is manually specified by a user of the system 100.
[058] In an embodiment, the reference database is generated using a corpus based sentiment analysis on data comprising semantics of a plurality of words from one or more languages, in a plurality of iterations, wherein in each of the plurality of iterations, associations between the plurality of words is established and updated, wherein a lexicon based dynamic crawling method is used to determine sentiment of words during the sentiment analysis.
[059] Further, at step 304 of the method 300, the system 100 segregates the data as one or more of a text component, an audio/video (av) component, an infographics component, and a currency component. Depending on the data in the document, the document may have all these components or a combination of one or more of these components.
[060] If the document has a text component, at step 306 of the method 300, the system 100 processes the text component using a text translation technique, to convert the data in the text component from the current language to the target language. In an embodiment, processing of the text component using the text translation technique includes translation of text, infographics, phases, Parts of Speech (PoS), and idioms, in the text component to the target language.
[061] Further, if the document has the a/v component, at step 308 of the method 300, the system 100 processes the a/v component using an a/v translation technique, to convert the data in the a/v component from the current language to the target language. Various steps involved in processing of the a/v component using the a/v translation technique are depicted in method 500 in FIG. 5,and are explained hereafter. At step 502 of the method 500, the system 100 captures modulation of language in the a/v component along with frequency of associated waves. Further, at step 504 of the method 500, a plurality of word contours (WC) and alphabet contours (AC) in the a/v component are identified, based on the captured modulation of language. Further, at step 506 of the method 500, one or more words matching the a/v component are identified by mapping the WCs and ACs to a plurality of pointers in a reference data. Further, at step 508 of the method 500, the identified one or more words are translated to the target language, based on a determined context of the document fetched as input and an associated emotion of a text being translated. For example, consider that the a/v component shows a person greeting another person by saying ‘Namaste’ and with closed hands, the Indian way of greeting people. While translating this a/v component, for example, to Japanese language, instead of providing literal translation of the term ‘Namaste’, the system 100 interprets that the context is greeting a person, and may show an emoji of bending down along with translated text corresponding to ‘Namaste’, which is the Japanese way of greeting people.
[062] Determining/identifying the context while translating the data involves determining a sentiment associated with each sentence being translated, by performing a sentiment analysis. The sentiment analysis is explained below.
[063] Performing the sentiment analysis by the system 100 involves classifying data point in each sentence as one of i) Positive (+1), ii) Negative (-1), and iii) Neither or Neutral (0). In order to determine sentiment of a word, the system 100 initially segregates verbs, conjunctions or connectors, as they do not have any sentiments. Consider the example in Table. 6.

Original Consumers Complain about the product which Is famous
Exclusion - - Excluded Excluded - - -

Table. 6

[064] Further, inputs being given to a word are being assigned Neutral (0) classification by default during the translation. This is given in Table. 7.

Step 1 Consumers Complain about the product which is famous
Exclusion - - Excluded Excluded - Excluded Excluded -
Sentiment_a 0 0 0 0
Table. 7
[065] Further, each word in the sentence, apart from the ones which are excluded, is subjected to check as per the sentiment of the input language. This is depicted in Table. 8.

Step 1 Consumers Complain about the product which is famous
Sentiment - - Excluded Excluded - Excluded Excluded -
Sentiment_a 0 0 0 0
Sentiment_b 0 -1 0 +1

Table. 8

[066] By means of the check, the system 100 determines that the whole sentence consists of 1 positive and 1 negative sentiment, which overall makes the sentiment of the aspect or the context neutral in nature. Further, the translated language sentiment is being checked, and various possible scenarios are explained below.
Case 1: – Sentiment of the target language is same as that of the current language (As per the stored/mapped lexicon in database). This is depicted in Table. 9.

Step 1 Consumers Complain about the product which is famous
Sentiment - - Excluded Excluded - Excluded Excluded -
Sentiment_a 0 0 0 0
Sentiment_b 0 -1 0 +1
Translated 0 -1 +1

Table. 9

Case 2 – Sentiment of the target language is different from that of the current language (As per the stored/mapped lexicon in database). This is depicted in Table. 9.

Step 1 Consumers Complain about the product which is famous
Sentiment - - Excluded Excluded - Excluded Excluded -
Sentiment_a 0 0 0 0
Sentiment_b 0 -1 0 +1
Translated 0 +1 +1

Table. 10

[067] If the sentiments are the same, then original context is retained. If the sentiments are different, then the specific words are subjected to further analysis using a corpus-based approach of sentiment analysis, where the semantics are further analyzed. Further, the words are subjected to different forms by applying decision-tree approach, along with using CMM (Conditional Markov Model) to determine a pre and the post linked word. This is depicted in Table. 11.

Step 1 Consumers Complain about the product which is famous
Sentiment - - Excluded Excluded - Excluded Excluded -
Sentiment_a 0 0 0 0
Sentiment_b 0 -1 0 +1
Translated 0 +1 +1
Complain about
To complain
Complaint for
-------
Complain against
Table. 11

[068] The sentiments of the phrases thus generated are being considered and maximum function is used in order to determine the sentiment of the related words or the total phrase. This is depicted in Table. 12.

Step 1 Consumers Complain about the product which is famous
Sentiment - - Excluded Excluded - Excluded Excluded -
Sentiment_a 0 0 0 0
Sentiment_b 0 -1 0 +1
Translated 0 +1 +1
Complain about
-1
To complain
-1
Complaint for
-1
-------
Complain against
-1

Table. 12

[069] The aforementioned steps of sentiment analysis are iterated till the sentiments of all the phrases generated in the target language is same as that of the current language. The semantic analysis coupled with word association is used to finalize the sentiment, as given in Table. 13.

Step 1 Consumers Complain about the product which is famous
Sentiment - - Excluded Excluded - Excluded Excluded -
Sentiment_a 0 0 0 0
Sentiment_b 0 -1 0 +1
Translated 0 +1 +1
Complain about
-1
To complain
-1
Complaint for
-1
-------
Complain against
-1
Final Sentiment 0 -1 0 +1

Table. 13

[070] The final sentiment is then retained by the system 100 for the target language and a literal translated word is also translated to the nearest meaning word having the finally determined sentiment. Further, selection of the linked word is done through the antonym-synonym lexicon/dictionary of linked words along with a secondary source search through a suitable approach (for example, a crawling approach) for finding the antonym and synonym from the secondary sources. After determining the word sentiments, the system 100 may determine line sentiments, followed by paragraph sentiment, and followed by a final aspect or a context sentiment. The sentiments are used for the contextualization of the text. In an embodiment, the sentences and determined sentiments and contextual data maybe used as training data to generate a data model, wherein the data model maybe then used to process sentences and determine associated sentiment.
[071] Referring back to method 300, if the document has an infographic component, at step 310 of the method 300, the system 100 processes the infographics component using an infographics translation technique, to convert the data in the infographics component from the current language to the target language. Various steps involved in processing the infographics component using the infographics translation technique are depicted in method 400 in FIG. 4, and are explained hereafter. At step 402 of the method 400, one or more infographics in the infographics component are scanned to determine a plurality of word contours (WC) and alphabet contours (AC). Further, at step 404 of the method 400, the plurality of WCs and ACs are segmented and classified as different PoS, to generate a plurality of segments. Further, at step 406 of the method 400, each of the plurality of segments is labelled and mapped to words present in each of a plurality of languages stored in a reference database, by the system 100. Further, at step 408 of the method 400, one or more words matching the one or more infographics is identified by comparing each of a plurality of combinations of the plurality of segments and the words with a plurality of reference data pointers. Further, at step 410 of the method 400, the identified one or more words are translated to the target language, based on a determined context of the document fetched as input and an associated emotion of a text being translated, wherein the associated emotion is determined in terms of one or more cultural proximities of the current language. The example given for a/v translation is applicable here as well. For example, consider that the infographics component shows a person greeting another person with closed hands, the Indian way of greeting people. While translating this infographics component, for example, to Japanese language, the system 100 interprets that the context is greeting a person, and may show an emoji of bending down in the Japanese way of greeting people, which is more customized to match the cultural proximities of the current language.
[072] Referring back to the method 300, if the document has the currency component, at step 312 of the method 300, the system 100 processes the currency component using a currency translation technique, to convert the data in the currency component from the current language to the target language. Various steps in the process of translating the currency component using the currency translation technique are depicted in method 600 in FIG. 6,and are explained hereafter. At step 602 of the method 600, the system 100 determines a currency type and a currency value, wherein the currency type is determined based on a) a currency type specified adjacent to the currency value, or b) based on the determined current language of the document. For example, if the text in the document is “100 INR”, the currency type is Indian Rupee. In another example, if the narration of content in the document is Indian context, then the system 100 may determine the currency type as Indian Rupee, even if a symbol representing the currency type is not given adjacent to currency value in the document. In order to ensure that latest conversion rate has been applied while performing the currency conversion, at the time of the currency conversion, at step 604 of the method 600, one of a) a current date, b) a date of compilation of the document, or c) a selected date, is determined as baseline for currency conversion, based on a received user input. Further, at step 606 of the method 600, a currency conversion rate with respect to the determined baseline is determined. Further, at step 608 of the method 600, the currency conversion is performed based on the currency value and the determined currency conversion rate.
[073] In an embodiment, the language translation by the system 100 maybe used for Image/Audio/ Video to Text/speech Translation for differently abled users. In this scenario, all the information in a translated output may be in the form of text or GIFs, or images or infographics, video, audio, and other formats, as maybe opted by the user. For example, a user who is deaf may opt for output in visual format (i.e. GIFs, images or infographics, video with subtitle, or sign language format), whereas a user who is blind may opt for output in audio format and/or Braille Language format. For example, "How can images be made auditory - Like a description of an image." - " We see a WHITE BOAT on A BLUE VAST LAKE on a BRIGHT SUNNY DAY. TORN WHITE CLOUD and GREEN HORIZON ANIMATED by Gentle BREEZE" - language translations of the IMAGES being very visual so that it appeals. This is followed by the audio in that whole scene.
[074] The written description describes the subject matter herein to enable any person skilled in the art to make and use the embodiments. The scope of the subject matter embodiments is defined by the claims and may include other modifications that occur to those skilled in the art. Such other modifications are intended to be within the scope of the claims if they have similar elements that do not differ from the literal language of the claims or if they include equivalent elements with insubstantial differences from the literal language of the claims.
[075] The embodiments of present disclosure herein address unresolved problem of context specific language translation. The embodiment, thus provides a mechanism of processing a document to translate different types of data in the document, till a pre-determined confidence score is obtained.
[076] It is to be understood that the scope of the protection is extended to such a program and in addition to a computer-readable means having a message therein; such computer-readable storage means contain program-code means for implementation of one or more steps of the method, when the program runs on a server or mobile device or any suitable programmable device. The hardware device can be any kind of device which can be programmed including e.g., any kind of computer like a server or a personal computer, or the like, or any combination thereof. The device may also include means which could be e.g., hardware means like e.g., an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or a combination of hardware and software means, e.g., an ASIC and an FPGA, or at least one microprocessor and at least one memory with software processing components located therein. Thus, the means can include both hardware means and software means. The method embodiments described herein could be implemented in hardware and software. The device may also include software means. Alternatively, the embodiments may be implemented on different hardware devices, e.g., using a plurality of CPUs.
[077] The embodiments herein can comprise hardware and software elements. The embodiments that are implemented in software include but are not limited to, firmware, resident software, microcode, etc. The functions performed by various components described herein may be implemented in other components or combinations of other components. For the purposes of this description, a computer-usable or computer readable medium can be any apparatus that can comprise, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device.
[078] The illustrated steps are set out to explain the exemplary embodiments shown, and it should be anticipated that ongoing technological development will change the manner in which particular functions are performed. These examples are presented herein for purposes of illustration, and not limitation. Further, the boundaries of the functional building blocks have been arbitrarily defined herein for the convenience of the description. Alternative boundaries can be defined so long as the specified functions and relationships thereof are appropriately performed. Alternatives (including equivalents, extensions, variations, deviations, etc., of those described herein) will be apparent to persons skilled in the relevant art(s) based on the teachings contained herein. Such alternatives fall within the scope of the disclosed embodiments. Also, the words “comprising,” “having,” “containing,” and “including,” and other similar forms are intended to be equivalent in meaning and be open ended in that an item or items following any one of these words is not meant to be an exhaustive listing of such item or items, or meant to be limited to only the listed item or items. It must also be noted that as used herein and in the appended claims, the singular forms “a,” “an,” and “the” include plural references unless the context clearly dictates otherwise.
[079] Furthermore, one or more computer-readable storage media may be utilized in implementing embodiments consistent with the present disclosure. A computer-readable storage medium refers to any type of physical memory on which information or data readable by a processor may be stored. Thus, a computer-readable storage medium may store instructions for execution by one or more processors, including instructions for causing the processor(s) to perform steps or stages consistent with the embodiments described herein. The term “computer-readable medium” should be understood to include tangible items and exclude carrier waves and transient signals, i.e., be non-transitory. Examples include random access memory (RAM), read-only memory (ROM), volatile memory, nonvolatile memory, hard drives, CD ROMs, DVDs, flash drives, disks, and any other known physical storage media.
[080] It is intended that the disclosure and examples be considered as exemplary only, with a true scope of disclosed embodiments being indicated by the following claims.
, Claims:We Claim:
1. A processor implemented method (200), comprising:
fetching (202), via one or more hardware processors, a document as input;
determining (204), via the one or more hardware processors, one or more data types in the document, wherein by determining the one or more data types in the document, data in the document is segregated as data that can be processed and data that cannot be processed; and
translating (206) the data that has been segregated as the data that can be processed, using a data processing model, via the one or more hardware processors, comprising, performing till a measured confidence score at least matches a threshold of confidence score:
determining (302) a current language of the document and a target language;
segregating (304) the data as one or more of a text component, an audio/video (av) component, an infographics component, and a currency component;
processing (306) the text component using a text translation technique, to convert the data in the text component from the current language to the target language;
processing (308) the a/v component using an a/v translation technique, to convert the data in the a/v component from the current language to the target language;
processing (310) the infographics component using an infographics translation technique, to convert the data in the infographics component from the current language to the target language; and
processing (312) the currency component using a currency translation technique, to convert the data in the currency component from the current language to the target language.

2. The method as claimed in claim 1, wherein processing the text component using the text translation technique comprises translation of text, infographics, phases, Parts of Speech (PoS), and idioms, in the text component to the target language.

3. The method as claimed in claim 1, wherein processing the infographics component using the infographics translation technique comprises:
scanning (402) one or more infographics in the infographics component to determine a plurality of word contours (WC) and alphabet contours (AC);
segmenting (404) and classifying the plurality of WCs and ACs as different PoS, to generate a plurality of segments;
labelling (406) and mapping each of the plurality of segments to words present in each of a plurality of languages stored in a reference database;
identifying (408) one or more words matching the one or more infographics by comparing each of a plurality of combinations of the plurality of segments and the words with a plurality of reference data pointers; and
translating (410) the identified one or more words to the target language, based on a determined context of the document fetched as input and an associated emotion of a text being translated, wherein the associated emotion is determined in terms of one or more cultural proximities of the current language.

4. The method as claimed in claim 3, wherein the reference database is generated using a corpus based sentiment analysis on data comprising semantics of a plurality of words from one or more languages, in a plurality of iterations, wherein in each of the plurality of iterations, associations between the plurality of words is established and updated, wherein a lexicon based dynamic crawling method is used to determine sentiment of words during the sentiment analysis.

5. The method as claimed in claim 1, wherein processing the a/v component using the a/v translation technique comprises:
capturing (502) modulation of language in the a/v component along with frequency of associated waves;
determine (504) a plurality of word contours (WC) and alphabet contours (AC) in the a/v component, based on the captured modulation of language;
identifying (506) one or more words matching the a/v component, by mapping the WCs and ACs to a plurality of pointers in a reference data; and
translating (508) the identified one or more words to the target language, based on a determined context of the document fetched as input and an associated emotion of a text being translated.

6. The method as claimed in claim 5, wherein translating the identified one or more words to the target language comprises generating a digital representation of the translation.

7. The method as claimed in claim 1, wherein processing the currency component using the currency translation technique comprises:
determining (602) by processing the document, a currency type and a currency value, wherein the currency type is determined based on a) a currency type specified adjacent to the currency value, or b) based on the determined current language of the document;
determining (604), based on a received user input, one of a) a current date, b) a date of compilation of the document, or c) a selected date, as baseline for currency conversion;
determining (606) a currency conversion rate with respect to the determined baseline; and
performing (608) currency conversion based on the currency value and the determined currency conversion rate.

8. The method as claimed in claim 1, wherein the current language is determined by processing the document fetched as input, comprising:
comparing (702) a plurality of words from the document with words in a plurality of languages in a reference database, wherein the plurality of words is selected from a portion of the document identified using an adaptive progress language technique;
determining (704) a similarity percentage for each of the plurality of languages, based on extent of similarity of the plurality of words with each of the plurality of languages; and
selecting (706) a language from among the plurality of languages, having the similarity percentage exceeding a threshold of similarity, as the current language.

9. A system (100), comprising:
one or more hardware processors (102);
a communication interface (112); and
a memory (104) storing a plurality of instructions, wherein the plurality of instructions when executed, cause the one or more hardware processors to:
fetch a document as input;
determine one or more data types in the document, wherein by determining the one or more data types in the document, data in the document is segregated as data that can be processed and data that cannot be processed; and
translate the data that has been segregated as the data that can be processed, using a data processing model, by performing till a measured confidence score at least matches a threshold of confidence score:
determining a current language of the document and a target language;
segregating the data as one or more of a text component, an audio/video (av) component, an infographics component, and a currency component;
processing the text component using a text translation technique, to convert the data in the text component from the current language to the target language;
processing the a/v component using an a/v translation technique, to convert the data in the a/v component from the current language to the target language;
processing the infographics component using an infographics translation technique, to convert the data in the infographics component from the current language to the target language; and
processing the currency component using a currency translation technique, to convert the data in the currency component from the current language to the target language.

10. The system as claimed in claim 9, wherein the one or more hardware processors are configured to translate text, infographics, phases, Parts of Speech (PoS), and idioms, in the text component to the target language, by processing the text component using the text translation technique.

11. The system as claimed in claim 9, wherein the one or more hardware processors are configured to process the infographics component using the infographics translation technique by:
scanning one or more infographics in the infographics component to determine a plurality of word contours (WC) and alphabet contours (AC);
segmenting and classifying the plurality of WCs and ACs as different PoS, to generate a plurality of segments;
labelling and mapping each of the plurality of segments to words present in each of a plurality of languages stored in a reference database;
identifying one or more words matching the one or more infographics by comparing each of a plurality of combinations of the plurality of segments and the words with a plurality of reference data pointers; and
translating the identified one or more words to the target language, based on a determined context of the document fetched as input and an associated emotion of a text being translated, wherein the associated emotion is determined in terms of one or more cultural proximities of the current language.

12. The system as claimed in claim 11, wherein the one or more hardware processors are configured to generate the reference database using a corpus based sentiment analysis on data comprising semantics of a plurality of words from one or more languages, in a plurality of iterations, wherein in each of the plurality of iterations, associations between the plurality of words is established and updated, wherein a lexicon based dynamic crawling method is used to determine sentiment of words during the sentiment analysis.

13. The system as claimed in claim 9, wherein the one or more hardware processors are configured to process the a/v component using the a/v translation technique by:
capturing modulation of language in the a/v component along with frequency of associated waves;
determine a plurality of word contours (WC) and alphabet contours (AC) in the a/v component, based on the captured modulation of language;
identifying one or more words matching the a/v component, by mapping the WCs and ACs to a plurality of pointers in a reference data; and
translating the identified one or more words to the target language, based on a determined context of the document fetched as input and an associated emotion of a text being translated.

14. The system as claimed in claim 13, wherein the one or more hardware processors are configured to translate the identified one or more words to the target language by generating a digital representation of the translation.

15. The system as claimed in claim 9, wherein the one or more hardware processors are configured to process the currency component using the currency translation technique by:
determining by processing the document, a currency type and a currency value, wherein the currency type is determined based on a) a currency type specified adjacent to the currency value, or b) based on the determined current language of the document;
determining, based on a received user input, one of a) a current date, b) a date of compilation of the document, or c) a selected date, as baseline for currency conversion;
determining a currency conversion rate with respect to the determined baseline; and
performing currency conversion based on the currency value and the determined currency conversion rate.

16. The system as claimed in claim 9, wherein the one or more hardware processors are configured to determine the current language by processing the document fetched as input, comprising:
comparing a plurality of words from the document with words in a plurality of languages in a reference database, wherein the plurality of words is selected from a portion of the document identified using an adaptive progress language technique;
determining a similarity percentage for each of the plurality of languages, based on extent of similarity of the plurality of words with each of the plurality of languages; and
selecting a language from among the plurality of languages, having the similarity percentage exceeding a threshold of similarity, as the current language.

Dated this 6th Day of February 2023

Tata Consultancy Services Limited
By their Agent & Attorney

(Adheesh Nargolkar)
of Khaitan & Co
Reg No IN-PA-1086