aSYSTEM AND METHOD TO ANALYSE AND PREDICT IMPACT OF TEXTUAL DATA
FIELD OF INVENTION
[0001] Embodiments of the present disclosure relate to data prediction, and more
particularly to system and method to analyse and predict impact of textual data.
BACKGROUND
[0002] Natural Language Processing is a type of analysis method in Machine
Learning or Artificial Intelligence which deals with extracting information from unstructured data and using extracted data to analyze and predict various trends and behavioural pattern and its influence on business performance.
[0003] In one approach, the system includes a processor configured to identify a
process to analyse data received from a source. Moreover, the data received from the source is a type of unstructured data which is in a form of natural language. Furthermore, the system searches and matches a Natural Language Processing (NLP) framework to predict and analyse the data received. However, the data received from the source could be either internal or external. Also, the data received does not depend on a use case or a specific context which lowers an efficiency of the system. Furthermore, the system requires a special set of instructions to analyse and predict the data, which involves high dependency on manual task from a user. Such dependency causes another factor to lower the efficiency of the system and also slows down the system. Moreover, in such system, the user must correlate the derived insights obtained during analysis of the data with the key performance indication. Such correlation by the user is yet another factor for less efficiency and delay of the system.
[0004] Hence, there is a need for an improved system and method to analyse and
predict impact of textual data to address the aforementioned issue.
BRIEF DESCRIPTION

[0005] In accordance with an embodiment of the present disclosure, a system to
analyse and predict impact of textual data is provided. The system includes a memory. The memory is configured to store a plurality of data sets acquired from one or more sources. The system also includes a processing subsystem operatively coupled to the memory. The processing subsystem is configured to select textual data from the plurality of data sets. The processing subsystem is also configured to extract data from one or more external sources through web crawling. The processing subsystem is also configured to identify at least one context of the textual data using one or more identification methods. The processing subsystem includes a natural language processing (NLP) module. The NLP module is configured to match the textual data with at least one natural language processing (NLP) frameworks using a mapping method based on a plurality of parameters. The NLP module is also configured to apply feature engineering and transformation on the textual data to extract a plurality of features from the plurality of data sets. The NLP module is also configured to analyse matched textual data of the textual using at least one analysis method, wherein the at least one analysis method includes at least one of a part of speech (POS) tagging, a sentiment method, a topic modelling, a clustering method and a document classification method. The NLP module is also configured to store an analysed result of the textual data in the memory. The processing subsystem also includes a predictive module. The predictive module is configured to obtain the analysed result of the textual data from the memory. The predictive module is also configured to predict one or more future values of the analysed textual data using the one or more predictive methods based on an analysis result.
[0006] In accordance with another embodiment, a method for analysing and
predicting impact of textual data is provided. The method includes acquiring a plurality of data sets from one or more sources. The method also includes selecting textual data from the plurality of data sets. The method also includes identifying at least one context of the textual data using one or more identification methods. The method also includes matching the textual data with at least one natural language processing (NLP) frameworks using a mapping method based on a plurality of parameters. The method also includes applying feature engineering and transformation on the textual data to extract a plurality of features from the plurality of data sets. The method also includes analysing matched textual data using at least one analysis

method. The method also includes predicting one or more future values of the analysed textual data using the one or more machine learning models based on an analysis result.
[0007] To further clarify the advantages and features of the present disclosure, a
more particular description of the disclosure will follow by reference to specific embodiments thereof, which are illustrated in the appended figures. It is to be appreciated that these figures depict only typical embodiments of the disclosure and are therefore not to be considered limiting in scope. The disclosure will be described and explained with additional specificity and detail with the appended figures.
BRIEF DESCRIPTION OF THE DRAWINGS
The disclosure will be described and explained with additional specificity and detail with the accompanying figures in which:
[0008] FIG. 1 is a block diagram representation of a system to analyse and predict
impact of textual data in accordance with an embodiment of the present disclosure;
[0009] FIG. 2 is a schematic representation of an exemplary embodiment of the
system to analyse and predict impact of the textual data of FIG. 1 in accordance with an embodiment of the present disclosure;
[0010] FIG. 3 is a block diagram representation of an exemplary embodiment of a
system to analyse and predict impact of textual data associated to an article of FIG. 1 in accordance with an embodiment of the present disclosure; and
[0011] FIG. 4 is a flow chart representing steps involved in a method for analysing
and predicting impact of textual data in accordance with an embodiment of the present disclosure.
[0012] Further, those skilled in the art will appreciate that elements in the figures
are illustrated for simplicity and may not have necessarily been drawn to scale. Furthermore, in terms of the construction of the device, one or more components of the device may have been represented in the figures by conventional symbols, and the figures may show only those specific details that are pertinent to understanding the embodiments of the present disclosure so as not to obscure the figures with details that

will be readily apparent to those skilled in the art having the benefit of the description herein.
DETAILED DESCRIPTION
[0013] For the purpose of promoting an understanding of the principles of the
disclosure, reference will now be made to the embodiment illustrated in the figures and specific language will be used to describe them. It will nevertheless be understood that no limitation of the scope of the disclosure is thereby intended. Such alterations and further modifications in the illustrated system, and such further applications of the principles of the disclosure as would normally occur to those skilled in the art are to be construed as being within the scope of the present disclosure.
[0014] The terms "comprises", "comprising", or any other variations thereof, are
intended to cover a non-exclusive inclusion, such that a process or method that comprises a list of steps does not include only those steps but may include other steps not expressly listed or inherent to such a process or method. Similarly, one or more devices or sub-systems or elements or structures or components preceded by "comprises... a" does not, without more constraints, preclude the existence of other devices, sub-systems, elements, structures, components, additional devices, additional sub-systems, additional elements, additional structures or additional components. Appearances of the phrase "in an embodiment", "in another embodiment" and similar language throughout this specification may, but not necessarily do, all refer to the same embodiment.
[0015] Unless otherwise defined, all technical and scientific terms used herein have
the same meaning as commonly understood by those skilled in the art to which this disclosure belongs. The system, methods, and examples provided herein are only illustrative and not intended to be limiting.
[0016] In the following specification and the claims, reference will be made to a
number of terms, which shall be defined to have the following meanings. The singular forms “a”, “an”, and “the” include plural references unless the context clearly dictates otherwise.

[0017] Embodiments of the present disclosure relates to system and method to
analyse and predict impact of textual data. The system includes a memory. The memory is configured to store a plurality of data sets acquired from one or more sources. The system also includes a processing subsystem operatively coupled to the memory. The processing subsystem is configured to select textual data from the plurality of data sets. The processing subsystem is also configured to extract data from one or more external sources through web crawling The processing subsystem is also configured to identify at least one context of the textual data using one or more identification methods. The processing subsystem includes a natural language processing (NLP) module. The NLP module is configured to match the textual data with at least one natural language processing (NLP) frameworks using a mapping method based on a plurality of parameters. The NLP module is also configured to apply feature engineering and transformation on the textual data to extract a plurality of features from the plurality of data sets. The NLP module is also configured to analyse matched textual data of the textual using at least one analysis method, wherein the at least one analysis method includes at least one of a part of speech (POS) tagging, a sentiment method, a topic modelling, a clustering method and a document classification method. The NLP module is also configured to store an analysed result of the textual data in the memory. The processing subsystem also includes a predictive module. The predictive module is configured to obtain the analysed result of the textual data from the memory. The predictive module is also configured to predict one or more future values of the analysed textual data using the one or more predictive methods based on an analysis result.
[0018] FIG. 1 is a block diagram (10) representation of a system (15) to analyse
and predict impact of textual data in accordance with an embodiment of the present disclosure. The system (15) includes a memory (20). The memory (20) is configured to store a plurality of data sets acquired from one or more sources (30). In one embodiment, the memory (20) may be a random-access memory (RAM), a read only memory (ROM), a cache memory or a flash memory. In one exemplary embodiment, the plurality of data sets may be acquired from at least one of a web, a manual entry of data, a local data set, an internal storage, an external storage and an experimental data set. In such embodiment, the plurality of data sets may include a plurality of structured data, a plurality of unstructured data or a plurality of semi-structured data.

In one exemplary embodiment, the plurality of data sets may include textual data or categorical data. In one specific embodiment, the memory (20) may correspond to a database of the system (15).
[0019] The system (15) also includes a processing subsystem (40) operatively
coupled to the memory (20). The processing subsystem (40) is configured to select textual data from the plurality of data sets. In one embodiment, the textual data may be selected by a user from the plurality of data sets. In another embodiment, the processing subsystem (40) may select the textual data from the plurality of data sets based on a pre-defined set of instructions. In yet another embodiment, the textual data may be selected by the processing subsystem (40) from at least one of the use case, the statistical influence and the previous predictive sample. The processing subsystem is also configured to extract data from one or more external sources through web crawling. As used herein, the term ‘web crawling’ is defined as a process for browsing the web to extract the data from the one or more external sources.
[0020] Furthermore, the processing subsystem (40) is also configured to identify at
least one context of the textual data using one or more identification methods. In one embodiment, the one or more identification methods may include one or more machine learning models. As used herein, the machine learning model is defined as a model built within a computer system using artificial intelligence which often adopts statistical techniques and computational learning theory in order to perform at least one said task. In one embodiment, the processing subsystem (40) may identify the at least one context from the built corpus using the one or more machine learning models.
[0021] The processing subsystem (40) includes a natural language processing
(NLP) module (42). The NLP module (42) is configured to match the textual data with at least one natural language processing (NLP) framework using a mapping method based on a plurality of parameters. In one embodiment, the mapping method may correspond to an artificial intelligence technique. As used herein, the artificial intelligence technique is defined as a type of intelligence demonstrate by a machine which is in contrast to the natural intelligence demonstrated by humans. Furthermore, in one exemplary embodiment, if the artificial intelligence technique fails to match the textual data with an appropriate at least one NPL framework which may be stored in the database corresponding to the memory (20), the artificial intelligence technique

may be used by the processing subsystem (40) to search the one or more external sources in order to find the appropriate at least one NPL framework which matches the textual data. Furthermore, if the user is not satisfied from the at least one NPL framework found from the one or more external sources, the user may create a new enterprise framework based on user’s preferences. In one embodiment, the plurality of parameters may include the use case, the statistical influence and the previous predictive sample.
[0022] The NLP module (42) is also configured to apply feature engineering and
transformation on the textual data to extract a plurality of features from the plurality of data sets. As used herein, feature engineering is the process of using domain knowledge of the data to create features that make one or more machine learning models work.
[0023] The NLP module (42) is also configured to analyse matched textual data
using at least one analysis method. Furthermore, the at least one analysis method includes at least one of a part of speech (POS) tagging, a sentiment method, a topic modelling, a clustering method and a document classification method. The NLP module (42) is also configured to store an analysed result of the textual data in the memory (20).
[0024] In one exemplary embodiment, upon selecting the at least one NPL
framework, the textual data of the textual data may be split into a training data model and a test data mode. Further, the processing subsystem (40) may use the training data model for self-learning. Upon self-learning of the training data model, the processing subsystem (40) tests the training data model for accuracy of the same. Furthermore, if the training data model matches an accuracy criterion, the built data model may be analysed using the at least one analysis method.
[0025] The processing subsystem (40) also includes a predictive module (44)
operatively coupled to the natural language processing (NLP) module (42). The predictive module (44) is configured to obtain the analysed result from the memory. The predictive module (44) is also configured to predict one or more future values of the analysed textual data using the one or more predictive methods based on the analysed result. In one embodiment, the processing subsystem (40) may predict the

one or more future values of a sentiment score which may be generated based on the NPL analysis method.
[0026] In one further embodiment, the system (15) may include a visualisation
engine (not shown in FIG. 1) operatively coupled to the processing subsystem (40). In one embodiment, the one or more predicted future values may be converted into dashboards and data visualizations by the visualisation engine. Further, a plurality of narratives for the visualisation of the may be generated automatically. The one or more predicted values may be displayed in a form of a story. In such embodiment, the user may store the analysis result and the insights in the memory. Further, the user may choose an appropriate KPI for correlation.
[0027] In another embodiment, the system (15) may include a representation
module operatively coupled to the visualisation engine. The representation module (not shown in FIG. 1) may be configured to represent the one or more predicted future values or the analysis result in one or more forms. In such embodiment, the one or more predicted future values or the analysis result may be represented in at least one of a graph, a chart, a table or an insight.
[0028] FIG. 2 is a schematic representation of an exemplary embodiment of the
system (50) to analyse and predict impact of the textual data of FIG. 1 in accordance with an embodiment of the present disclosure. A memory (not shown in FIG. 2) of the system (50) may mash up one or more external data sets (60) and one or more internal data sets (70) to generate the plurality of data sets using a mashup module (80). The one or more external data sets (60) is obtained from a web crawler (90). Further, the one or more internal data sets (70) is obtained from a plurality of already existing data set or a plurality of previously verified data set which may be present in the memory. Further, the plurality of data sets may be transmitted to a processing subsystem (not shown in FIG. 2) to pre-process the plurality of data sets. On processing the plurality of data sets, textual data may be selected by the processing subsystem. The system (50) includes a natural language processing (NLP) engine (100) located within the processing subsystem.
[0029] Furthermore, the textual data selected by the processing subsystem is
subjected to a corpus build module (110) located within the NLP engine (100) and

configured to build a corpus. Furthermore, on building the corpus, the processing subsystem identifies context of the built corpus. Consequently, the built corpus is mapped with an appropriate NLP model by a model mapper (120), wherein the model mapper (120) is located within the NLP engine (100).
[0030] On mapping the built corpus with the NLP model, the NLP engine (100)
identifies one or more feature engineering and transformations required to predict the built corpus. Furthermore, the selected one or more feature engineering and the transformations are run on the built corpus to extract a plurality of features. The built corpus is now transmitted to an NLP analysis module (130) which is located within the NLP engine (100). The NLP analysis module (130) identifies one or more analysis method to analyse the built corpus. Also, the processing subsystem selects an appropriate analysis method for regression. A prediction engine (not shown) also trains and testes the built corpus in order to predict one or more future values of the built corpus.
[0031] In addition, predicted result is passed to a data visualisation engine (140).
The data visualisation engine (140) generates one or more of a chart, a graph or a table based on the predicted result. The processing subsystem further generates a model summery based on the predicted results. Also, the predicted result along with generated insights are stored in the memory. Furthermore, the predicted result is displayed on a display (150) which is operatively coupled to the processing subsystem.
[0032] Furthermore, the system (50) which includes the memory and the
processing subsystem of FIG. 2 is substantially similar to a system (15) which includes a memory (20) and a processing subsystem (40) of FIG. 1.
[0033] FIG. 3 is a block diagram (160) representation of an exemplary embodiment
of a system (165) to analyse and predict impact of textual data associated to an article of FIG. 1 in accordance with an embodiment of the present disclosure. A user (170) uploads the article to a memory (180) of the system (165) through an external storage (190) means wherein the article is in an unstructured data format. The uploaded article is stored in the memory (180). Further, a plurality of data sets related to the article is acquired from one or more other sources (200) such as the external storage source (190) and an internal storage source (210). Upon acquiring the plurality of data sets

from different sources, the plurality of data sets is mashed and is stored in the memory (180) for further analysis and prediction of sentiment and thoughts related to the article.
[0034] Furthermore, as the user (170) selects the article for further analysis and
prediction, a processing subsystem (220) builds a corpus by selecting textual data from the uploaded article. Consequently, the processing subsystem (220) identifies context of the built corpus using a first machine learning model to correctly match the textual data to an appropriate natural language processing (NLP) framework for further analysis. Furthermore, the processing subsystem (220) identifies a right NLP framework based on the context, use case and nature of the built corpus related to the article using an artificial intelligence technique.
[0035] Upon identifying the right NLP framework, the processing subsystem (220)
applies a POS tagging, a sentiment technique, a topic modelling, a clustering method and a document classification technique on the built corpus to analyse the article. Further, based on the analyses done on the built corpus, the processing subsystem (220) analyses a sentiment and a future value of the article. The analysed sentiment value and the analysed future value is stored in the memory (180).
[0036] Also, a statistical model chooses the right technique to select a right second
machine learning model based on a data quality, data volume, computational resources required, data type, the use case, historical model performance. Upon selecting the right second machine learning model, the processing subsystem (220) splits the built corpus into a training model and a test model. The processing subsystem (220) uses the training model for self-learning. Consequently, the processing subsystem (220) tests the training model for accuracy with respect to a pre-defined accuracy rate.
[0037] Furthermore, based on the accuracy rate of the built corpus, the built corpus
is subjected to prediction of the analysed sentiment and the analysed future value using the second machine learning model. In addition, the processing subsystem (220) determines an impact of the analysed sentiment using the second machine learning model.
[0038] The predicted result is converted into dashboards and data visualisations by
the processing subsystem (220). In addition, a plurality of narratives for the

visualisation of the predicted result is also generated by the processing subsystem (220). The predicted result is displayed in form of stories on a display (230) which is operatively coupled to the processing subsystem (220). The stories are viewed by the user (170). Also, the user (170) chooses an enterprise database to store the analysed result and the insights which is operatively coupled to the memory (180). The user (170) also selects an appropriate key performance indicator (KPI) for correlation with the prediction of the article.
[0039] Furthermore, the sources (200), the memory (180) and the processing
subsystem (220) are substantially similar to sources (30), a memory (20) and a processing subsystem (40) od FIG. 1.
[0040] FIG. 4 is a flow chart representing steps involved in a method for analysing
and predicting impact of textual data in accordance with an embodiment of the present disclosure. The method (300) includes acquiring a plurality of data sets from one or more sources in step 310. In one embodiment, acquiring the plurality of data sets from the one or more sources may include acquiring the plurality of data sets from at least one of an internal storage and an external storage. In another embodiment, acquiring the plurality of data sets from the one or more sources may include acquiring the plurality of data sets from at least one of a web, a manual entry of data, a local data set, an internal storage, an external storage and an experimental data set.
[0041] The method (300) also includes selecting textual data from the plurality of
data sets in step320. In one embodiment, selecting the textual data from the plurality of data sets may include selecting the textual data based on a plurality of parameters such as a use case, a statistical influence and a previous predictive sample.
[0042] Furthermore, the method (300) includes identifying at least one context of
the textual data using one or more identification methods in step 340. In one embodiment, identifying the at least one context of the textual data may include identifying the at least one context of the textual data using one or more machine learning models.
[0043] The method (300) also includes matching the textual data with at least one
natural language processing (NLP) framework using a mapping method based on a plurality of parameters in step 350. In one embodiment, matching the textual data with

the at least one NLP framework may include matching the textual data with the at least one NLP framework from a stored database. In another embodiment, matching the textual data with the at least one NLP framework may include matching the textual data with the at least one NLP framework by searching the at least one NLP framework from one or more external open sources. In yet another embodiment, matching the textual data with the at least one NLP framework may include enabling a user to create the at least one framework based on one or more user preferences.
[0044] The method (300) also includes applying feature engineering and
transformation on the textual data to extract a plurality of features from the plurality of data sets in step 360. In one embodiment, applying feature engineering and transformation on the built textual data model may include applying the transformation on the built textual data model based on the use case, a data quality, a data type and a data volume for extracting the plurality of features from the plurality of data sets.
[0045] The method (300) also includes analysing matched textual data using at
least one analysis method in step 370. In one embodiment, analysing the matched textual data may include analysing the matched textual data of the textual data using at least one of a part of speech (POS) tagging, a sentiment method, a topic modelling, a clustering method and a document classification method.
[0046] The method (300) also includes predicting one or more future values of the
analysed textual data using the one or more predictive methods based on an analysis result in step 380. In one embodiment, predicting the one or more future values of the analysed textual data may include predicting one or more future values of the analysed textual data based on the types of analyses done on the textual data.
[0047] In one further embodiment, the method (300) may include representing one
or more predicted future values in one or more forms. In such embodiment, representing the one or more predicted future values in the one or more forms may include representing the one or more predicted future values in at least one of a graph, a chart, a table or an insight. In such another embodiment, representing the one or more predicted future values may include representing the one or more predicted future values in a form of a story.

[0048] In another embodiment, the method (300) may include storing the one or
more predicted future values or one or more insights in the memory. In such embodiment, storing the one or more predicted future values may include storing the one or more predicted future values in a database.
[0049] Various embodiments of the system and method to analyse and predict
impact of textual data enable the system to acquire the plurality of data sets from both the external source and the internal source. Furthermore, as the system uses the one or more machine learning models and the artificial intelligence technique, the system reduces the dependency of manual task from the user. Henceforth, increasing the efficiency of the system and decreasing the delay.
[0050] In addition, the system uses different parameters such as the use case, the
statistical influence and the previous predictive sample which adds on to the system to increase the efficiency of the prediction of the textual data and hence the efficiency of the system. The system also enables the user to analyse large scare unstructured data in an efficient and a faster method.
[0051] Also, the system automatically crawls the internal data and the external data
and builds the corpus automatically to analyse the textual data. In addition, the system also automatically adjusts the model parameters and chosen variables based on a feedback, wherein the feedback is the difference between the actual result and the predicted result. The system also automatically tracks and monitors the model performance.
[0052] While specific language has been used to describe the disclosure, any
limitations arising on account of the same are not intended. As would be apparent to a person skilled in the art, various working modifications may be made to the method in order to implement the inventive concept as taught herein.
[0053] The figures and the foregoing description give examples of embodiments.
Those skilled in the art will appreciate that one or more of the described elements may well be combined into a single functional element. Alternatively, certain elements may be split into multiple functional elements. Elements from one embodiment may be added to another embodiment. For example, order of processes described herein may be changed and are not limited to the manner described herein. Moreover, the actions

of any flow diagram need not be implemented in the order shown; nor do all of the acts need to be necessarily performed. Also, those acts that are not dependant on other acts may be performed in parallel with the other acts. The scope of embodiments is by no means limited by these specific examples.

WE CLAIM:
1. A system (15) to analyse and predict impact of textual data comprising:
a memory (20) configured to store a plurality of data sets acquired from one or more sources (30);
a processing subsystem (40) operatively coupled to the memory, and configured to:
select textual data from the plurality of data sets;
extract data from one or more external sources through web crawling;
identify at least one context of the textual data using one or more context identification methods,
wherein the processing subsystem comprises:
a natural language processing (NLP) module (42) configured to:
match the textual data with at least one natural language processing (NLP) framework from a plurality of frameworks obtained from the one or more sources (30) using a mapping method based on a plurality of parameters;
apply feature engineering and transformation on the textual data to extract a plurality of features from the plurality of data sets;
analyse matched textual data using at least one analysis method, wherein the at least one analysis method comprises at least one of a part of speech (POS) tagging, a sentiment method, a topic modelling, a clustering method and a document classification method;
store an analysed result of the textual data in the memory (20);
a predictive module (44) operatively coupled to the natural language processing (NLP) module (42), and configured to:

obtain the analysed result of the textual data from the memory (20); and
predict one or more future values of the analysed textual data using one or more predictive methods based on the analysed result.
2. The system (15) as claimed in claim 1, wherein the plurality of data sets comprises at least one of a plurality of structured data sets, a plurality of unstructured data sets and a plurality of semi-structured data sets.
3. The system (15) as claimed in claim 1, wherein the plurality of parameters comprises at least one of a use case, a statistical influence and a previous predictive sample.
4. The system (15) as claimed in claim 1, further comprises a representation module operatively coupled to the processing subsystem (40), and configured to represent one or more predicted future values in one or more forms.
5. A method (300) for analysing and predicting impact of textual data comprising:
acquiring a plurality of data sets from one or more sources; (310)
selecting textual data from the plurality of data sets; (320)
identifying at least one context of the textual data using one or more context identification methods; (330)
matching the textual data with at least one natural language processing (NLP) framework from a plurality of frameworks obtained from the one or more sources using a mapping method based on a plurality of parameters; (340)
applying feature engineering and transformation on the textual data to extract a plurality of features from the plurality of data sets; (350)
analysing matched textual data using at least one analysis method; and (360)
predicting one or more future values of the analysed textual data using the one or more predictive methods based on an analysis result. (370)

6. The method (300) as claimed in claim 5, wherein acquiring the plurality of data sets from one or more sources comprises acquiring the plurality of data from at least one of a web, a manual entry of data, a local data set, an internal storage, an external storage and an experimental data set.
7. The method (300) as claimed in claim 5, wherein analysing the matched textual data using the at least one analysis method comprises analysing the matched textual data using at least one of a part of speech (POS) tagging, a sentiment method, a topic modelling, a clustering method and a document classification method.
8. The method (300) as claimed in claim 5, further comprises representing one or more predicted future values of the textual data in one or more forms.