Description:
FIELD OF DISCLOSURE
[0001] Embodiments of the present invention relate to a method for generating responses in a conversational artificial intelligence (AI) platform, more specifically, relates to a method for generating personalized and faster responses in a conversational artificial intelligence (AI) platform.
BACKGROUND OF THE DISCLOSURE
[0002] In today’s digital age, the advent of e-commerce has transformed the interaction of the businesses with customers. Amidst this, conversational chatbots or platforms have emerged as a major technology, facilitating the convenience of instant messaging on various e-commerce platforms. Conversational chatbots or platforms enabling customers to engage with brands through chat interfaces, whether it be for making purchases, asking for recommendations, or seeking customer support.
[0003] With the widespread use of conversational chatbots or platforms have emerged the need for personalized and faster. In this digital age, the customers expect for immediate gratification and highly personalized interactions. With easy availability of information and services, consumers are becoming less willing to endure long wait times or generic responses.
[0004] Apart from the expectation of customers, the competitive landscape of e-commerce also necessitates that the businesses improve quality of their customer interactions. The businesses may benefit by providing a superior, personalized customer experience as efficient and personalized customer service can lead to higher conversion rates. Furthermore, swift and accurate responses can address a plurality of issues associated with cart abandonment rates, indecision or dissatisfaction of customers and lost sales.
[0005] Therefore, to support the dynamic nature of e-commerce, and extreme competitiveness in the field, necessitates that the conversational commerce platforms also integrate various advanced technologies to provide faster and more personalized responses to the customers. In modern times, with wide spread use of artificial intelligence and automation technologies in the conversational platforms, balancing between accuracy, speed, and personalization of response generated is a complex and challenging tasks.
[0006] The current conversational artificial intelligence (AI) platforms face challenges in delivering timely and personalized responses due to the need for processing complex queries in real-time. Current methods often involve high token utilization and latency issues, which can degrade user experience. There is a need for a system that can efficiently cache and retrieve user-specific data to enhance the performance of such platforms.
[0007] Thus, in light of the above-stated discussion, the present invention provides a method for generating personalized and faster responses in a conversational artificial intelligence (AI) platform by caching user intents and semantic data.
OBJECTIVES OF THE DISCLOSURE
[0008] The following is a summary description of illustrative embodiments of the invention. It is provided as a preface to assist those skilled in the art to more rapidly assimilate the detailed design discussion which ensues and is not intended in any way to limit the scope of the claims which are appended hereto in order to particularly point out the invention.
[0009] According to illustrative embodiments, the present disclosure focuses on the method and the system for caching user intents and semantic data for personalized responses in conversational artificial intelligence (AI)-platform, which overcomes the above-mentioned disadvantages or provide the users more personalized and faster responses.
[0010] The present disclosure solves all the above major limitations of a method and a system for generating faster and more personalized responses on a conversational artificial intelligence (AI) platform. Further, the present disclosure ensures that the disclosed invention may fulfil following objectives.
[0011] The principal object of this invention is to provide a method and system for efficient caching and retrieval of data for conversational artificial intelligence (AI) platform to generate more personalized and faster responses.
[0012] Another objective of this invention is to reduce latency and minimize the delay in generating responses to user queries and enhance the user experience by providing quicker answers, which is critical for maintaining user engagement and satisfaction in real-time interactions.
[0013] Yet another objective of this invention is to lower the number of tokens required for processing queries for faster processing times and lower operational costs.
[0014] Yet another objective of this invention is to develop robust caching strategies for storing frequently accessed data and common responses.
[0015] Yet another objective of this invention is to ensure responses are tailored to individual user preferences and history.
[0016] Yet another objective of this invention is to enable real-time data retrieval ensuring quick access to generate accurate and timely responses
[0017] Yet another objective of this invention is to improve user satisfaction, engagement, and retention, which are key metrics for the success of conversational artificial intelligence (AI) platforms.
SUMMARY OF THE DISCLOSURE
[0018] The following is a summary description of illustrative embodiments of the invention. It is provided as a preface to assist those skilled in the art to more rapidly assimilate the detailed design discussion which ensues and is not intended in any way to limit the scope of the claims which are appended hereto in order to particularly point out the invention.
[0019] In light of the above, in one aspect of the present disclosure, a method for caching user intents and semantic data for personalized responses in conversational artificial intelligence (AI)-platform is disclosed herein. The method includes receiving a user query through a messaging platform hosted on a user device, wherein a plurality of user query is handled simultaneously across a plurality of messaging platforms. The method also includes analysing the user query using a plurality of natural language processing (NLP) technique to recognize the user intent. The method also includes performing cache lookup, wherein the cache lookup involves checking a semantic data cache for pre-stored intents and associated data corresponding to the identified user intent. The method also includes storing the recognized intent in the semantic data cache along with associated semantic data and user profile information. The method also includes mapping the cached data with user attributes such as preferences, behaviour patterns, and demographic information. The method also includes retrieving relevant data from the semantic data cache and processing the user queries by fetching data based on pre-determined criteria. The method also includes generating a personalized response using the retrieved data.
[0020] In accordance with an embodiment of the present invention, the method performs token optimization by reducing the number of tokens required.
[0021] In accordance with an embodiment of the present invention, the use of pre-cached data reduces the number of tokens required for processing, and enhancing efficiency.
[0022] In accordance with an embodiment of the present invention, the method predicts future user intents based on historical interaction data.
[0023] In accordance with an embodiment of the present invention, the cache lookup further includes finding a relevant cached response based on the user query and attributes, if matching occurs, and forwarding the query to the artificial intelligence (AI)-backend infrastructure for processing, if no matching occurs.
[0024] In accordance with an embodiment of the present invention, the cache lookup also includes checking for pre-fetched data relevant to the current user query.
[0025] In accordance with an embodiment of the present invention, the method also includes visualization and reporting of key performance indicators (KPIs) through a plurality of dashboarding tool on a brand owner interface.
[0026] In another aspect of the present disclosure, a system for caching user intents and semantic data for personalized responses in conversational artificial intelligence (AI)-platform is disclosed herein. The system comprising a plurality of user device configured to allow a plurality of user to send a query through a plurality of messaging platform. The user device also provide response generated to the user. The system also comprising a user intent recognition module operationally coupled to the user device and the user intent recognition module configured to handle incoming user queries through natural language processing (NLP) to understand user intent. The user intent recognition module performs chat processing by performing a plurality of data operations including normalisation, context awareness, prompt engineering, and multi label classification. The system also comprising user profile database operationally coupled to the user intent recognition module and the user profile database configured to store and retrieve data related to the profile of the user. The system also comprising an artificial intelligence (AI)-backend infrastructure operationally coupled to the user intent recognition module and the artificial intelligence (AI)-backend infrastructure configured to leverage large language models (LLMs) and openAI to generate intelligent responses. The artificial intelligence (AI)-backend infrastructure includes a semantic data cache configured to cache recognized user intents and associated semantic data coupled with the user profile data to facilitate personalisation and faster response. The semantic data cache facilitates faster response generation by storing frequently accessed information. The system also comprising a data warehouse operationally coupled to the artificial intelligence (AI)-backend infrastructure and the data warehouse configured to aggregate data from the user profile database and other sources to provide a unified view for analysis. The system also comprising a structured query language (SQL) generator operationally coupled to the artificial intelligence (AI)-backend infrastructure and the structured query language (SQL) generator configured to generating structured query language (SQL) queries to retrieve specific information from the data warehouse. The structured query language (SQL) generator is integrated with the artificial intelligence (AI)-backend infrastructure to enhance query formulation and execution.
[0027] The system also comprising a band owner interface operationally coupled to the artificial intelligence (AI)-backend infrastructure and the band owner interface configured to provide insights and actionable data to improve customer engagement and sales strategies and allow brand owners to manage customer interactions and/ analyse performance metrics. The brand owner interface also includes a plurality of dashboarding tool for visualization and reporting of key performance indicators (KPIs).
[0028] In accordance with an embodiment of the present invention, the semantic data cache uses a cache hit counter operable to update cached data and remove outdated or irrelevant cached data.
[0029] In accordance with an embodiment of the present invention, the semantic data cache also maps the cached data with user specific attributes such as, preferences, behaviour patterns, and demographics.
[0030] In yet another embodiment of the present disclosure, a hybrid graph-based caching method for optimizing handling of the chatbot queries and responses is disclosed herein. The method includes constructing a dynamic graph structure where a plurality of node represents queries and user profiles, interconnected by a plurality of edge. The method also includes performing cache lookup/search by using a plurality of feature-based processes, employing a plurality of graph-based similarity measures, and thereafter performing cache indexing. The method also includes updating the cache by continuously updating the dynamic graph structure followed by edge weight update, and community detection update. The method also includes implementing eviction policies by simultaneously performing time-based expiration, usage-based eviction, and size-based eviction.
[0031] These and other advantages will be apparent from the present application of the embodiments described herein.
BRIEF DESCRIPTION OF THE DRAWINGS
[0032] To describe the technical solutions in the embodiments of the present disclosure or in the prior art more clearly, the following briefly describes the accompanying drawings required for describing the embodiments or the prior art. Apparently, the accompanying drawings in the following description merely show some embodiments of the present disclosure, and a person of ordinary skill in the art can derive other implementations from these accompanying drawings without creative efforts. All of the embodiments or the implementations shall fall within the protection scope of the present disclosure. The advantages and features of the present disclosure will become better understood with reference to the following detailed description taken in conjunction with the accompanying drawing, in which:
[0033] Fig, 1 illustrating a flowchart of a method for caching user intents and semantic data for personalized responses in conversational artificial intelligence (AI)-platform, in accordance with an exemplary embodiment of the present disclosure;
[0034] Fig. 2A illustrating a block diagram of a system for caching user intents and semantic data for personalized responses in conversational artificial intelligence (AI)-platform, in accordance with an exemplary embodiment of the present disclosure;
[0035] Fig. 2B illustrates a block diagram of an exemplary system architecture for the system for caching user intents and semantic data for personalized responses in conversational artificial intelligence (AI)-platform, in accordance with an exemplary embodiment of the present disclosure;
[0036] Fig. 3 illustrates a flowchart for a hybrid graph-based caching method for optimizing handling of the chatbot queries and responses, in accordance with an exemplary embodiment of the present disclosure; and
[0037] Fig. 4 illustrates an exemplary operational flow for the hybrid graph-based caching, in accordance with an exemplary embodiment of the present disclosure.
[0038] Like reference, numerals refer to like parts throughout the description of several views of the drawing. The method and system for caching user intents and semantic data for personalized responses in conversational artificial intelligence (AI)-platform, is illustrated in the accompanying drawings, which like reference letters indicate corresponding parts in the various figures. It should be noted that the accompanying figure is intended to present illustrations of exemplary embodiments of the present disclosure. This figure is not intended to limit the scope of the present disclosure. It should also be noted that the accompanying figure is not necessarily drawn to scale.
DETAILED DESCRIPTION OF THE DISCLOSURE
[0039] The following is a detailed description of embodiments of the disclosure depicted in the accompanying drawings. The embodiments are in such detail as to communicate the disclosure. However, the amount of detail offered is not intended to limit the anticipated variations of embodiments; on the contrary, the intention is to cover all modifications, equivalents, and alternatives falling within the spirit and scope of the present disclosure.
[0040] In the following description, numerous specific details are set forth in order to provide a thorough understanding of the embodiments of the present disclosure. It may be apparent to one skilled in the art that embodiments of the present disclosure may be practiced without some of these specific details.
[0041] The terms “a” and “an” herein do not denote a limitation of quantity, but rather denote the presence of at least one of the referenced items. The terms “having”, “comprising”, “including”, and variations thereof signify the presence of a component.
[0042] A conversational chatbot or a conversation platform is a computer program that simulates, as experienced by a user of an electronic communication device, such as a smartphone or laptop computer, as a conversation with a human being. The chatbot provides a bi-lateral conversation experience for the user, for example, a precise resolution of a customer query, or provisions for availing a requested service. Further, the chatbot can be a business engagement tool used in various business activities including, customer support, sale of a product or service, and such.
[0043] Systems and methods disclosed herein can provide, among other technical features and benefits that will be appreciated by persons of ordinary skill upon reading this disclosure, the capability to generate a personalized response to user query with reduced latency of response.
[0044] Referring now to Fig. 1 to Fig. 3B to describe various exemplary embodiments of the present disclosure. Fig. 1 illustrating a flowchart of a method 100 for caching user intents and semantic data for personalized responses in conversational artificial intelligence (AI)-platform, in accordance with an exemplary embodiment of the present disclosure. The method 100 may comprise the following steps.
[0045] At 102, receiving a user query through a messaging platform hosted on a user device 202. A plurality of user query is handled simultaneously across a plurality of messaging platforms.
[0046] In a preferred embodiment, the user device 202 may be compatible with acquiring both the textual input data and the voice input data from the user and transmit it downstream.
[0047] At 104, analysing the user query using a plurality of natural language processing (NLP) technique to recognize the user intent.
[0048] At 106, performing cache lookup. The cache lookup involves checking a semantic data cache 216 for pre-stored intents and associated data corresponding to the identified user intent.
[0049] At 108, storing the recognized intent in the semantic data cache 216 along with associated semantic data and user profile information.
[0050] At 110, mapping the cached data with user attributes such as preferences, behaviour patterns, and demographic information.
[0051] At 112, retrieving relevant data from the semantic data cache 216 and processing the user queries by fetching data based on pre-determined criteria.
[0052] At 114, generating a personalized response using the retrieved data.
[0053] The method 100 may perform token optimization by reducing the number of tokens required.
[0054] The use of pre-cached data may reduce the number of tokens required for processing, and enhancing efficiency.
[0055] The method 100 may predict future user intents based on historical interaction data.
[0056] The cache lookup may further include finding a relevant cached response based on the user query and attributes, if matching occurs and forwarding the query to an artificial intelligence (AI)-backend infrastructure 208 for processing, if no matching occurs. The cache lookup may also include checking for pre-fetched data relevant to the current user query.
[0057] The method 100 may also includes visualization and reporting of key performance indicators (KPIs) through a plurality of dashboarding tool on a brand owner interface 214.
[0058] In an embodiment of the present disclosure, the messaging platform may include, but not limited to, WhatsApp, LINE, and Facebook Messenger. The messaging platforms may allow the users to send their queries and receive responses. In an embodiment of the present disclosure, during chat processing the incoming user queries are pre-processed using a plurality of pre-processing operations.
[0059] In an embodiment of the present disclosure, the semantic data cache 216 may facilitate recognition of the user intents and associated semantic data and mapped it with user specific attributes. In an embodiment of the present disclosure, in case no matching of pre-stored user intent is found in the semantic data cache 216 and the query is forwarded to the artificial intelligence (AI)-backend infrastructure 208 for processing. The artificial intelligence (AI)-backend infrastructure 208 may use advanced language models such as, but not limited to, GenAI/LLM to processes the incoming user data and generate responses. Thereby, the final response generated is either by utilizing the cached data stored in the semantic data cache 216 or by the GenAI/LLM, which optimizes personalization and efficiency of the response generated.
[0060] In an embodiment of the present disclosure, after the response generation, the predictive model analyses historical interaction data and user attributes to predict future user intents by pre-fetching relevant responses to be ready for future queries. In a preferred embodiment, with the generation of the personalized responses by retrieving and utilizing cached data, the method 200 may reduce token utilization and reduce response time.
[0061] Fig. 2A illustrating a block diagram of a system 200 for caching user intents and semantic data for personalized responses in conversational artificial intelligence (AI)-platform, in accordance with an exemplary embodiment of the present disclosure.
[0062] The system 200 may comprise a plurality of user device 202, a user intent recognition module 204, a user profile database 206, an artificial intelligence (AI)-backend infrastructure 208 including a semantic data cache 216, a data warehouse 210, a structured query language (SQL) generator 212, and a band owner interface 214
[0063] The plurality of user device 202 may be configured to allow a plurality of user to send a query through a plurality of messaging platform. The user device 202 may also provide response generated to the user.
[0064] The user intent recognition module 204 may be operationally coupled to the user device 202 and the user intent recognition module 204 configured to handle incoming user queries through natural language processing (NLP) to understand user intent. The user intent recognition module 204 may perform chat processing by performing a plurality of data operations including normalisation, context awareness, prompt engineering, and multi label classification.
[0065] The user profile database 206 may be operationally coupled to the user intent recognition module 204 and the user profile database 206 configured to store and retrieve data related to the profile of the user.
[0066] The artificial intelligence (AI)-backend infrastructure 208 may be operationally coupled to the user intent recognition module 204 and the artificial intelligence (AI)-backend infrastructure 208 configured to leverage large language models (LLMs) and openAI to generate intelligent responses. The artificial intelligence (AI)-backend infrastructure 208 may include a semantic data cache 216 configured to cache recognized user intents and associated semantic data coupled with the user profile data to facilitate personalisation and faster response. The semantic data cache 216 may facilitate faster response generation by storing frequently accessed information.
[0067] The data warehouse 210 may be operationally coupled to the artificial intelligence (AI)-backend infrastructure 208 and the data warehouse 210 configured to aggregate data from the user profile database 206 and other sources to provide a unified view for analysis. The structured query language (SQL) generator 212 may be operationally coupled to the artificial intelligence (AI)-backend infrastructure 208 and the structured query language (SQL) generator 212 configured to generating structured query language (SQL) queries to retrieve specific information from the data warehouse. The structured query language (SQL) generator 212 may be integrated with the artificial intelligence (AI)-backend infrastructure 208 to enhance query formulation and execution.
[0068] The band owner interface 214 may be operationally coupled to the artificial intelligence (AI)-backend infrastructure 208 and the band owner interface 214 configured to provide insights and actionable data to improve customer engagement and sales strategies and allow brand owners to manage customer interactions and/ analyse performance metrics. The brand owner interface 214 may also include a plurality of dashboarding tool for visualization and reporting of key performance indicators (KPIs).
[0069] The semantic data cache 216 may use a cache hit counter operable to update cached data and remove outdated or irrelevant cached data.
[0070] The semantic data cache 216 may also map the cached data with user specific attributes such as, preferences, behaviour patterns, demographics.
[0071] In an embodiment of the present disclosure, the user device 202 may be a desktop computer, a laptop computer, a user computer, a tablet computer, a personal digital assistant (PDA), a cellular telephone, a smartphone, a media player, a navigation device, an email device, or a combination of any of the known, related art, or later developed data processing devices.
[0072] In some embodiments, when a user interacts with the conversational platform, the system (200) may predict the user's intent based on historical data and retrieves pre-cached information to generate a more personalized response. In an embodiment of the present disclosure, the user profile database 206 may contain detailed user-specific data such as, but not limited to, user preferences, behaviour patterns, and demographics, which is crucial for mapping intents and providing personalized responses. In an embodiment of the present disclosure, the user profile database 206 may use a MongoDB database to store and retrieve data.
[0073] In an embodiment of the present disclosure, the data warehouse 210 may utilize various data sources such as MongoDB, application data, and CSV files to provide the necessary background data for processing of the user query and personalization of the response generated. In a preferred embodiment, the data warehouse 210 may aggregate data from various sources to provide a unified view for analysis and feeds data to the structured query language (SQL) generator 212 and artificial intelligence (AI)-backend infrastructure 208 for processing.
[0074] In an embodiment of the present disclosure, the brand owner interface 214 may be integrated any electronic device, including but not limited to mobile phones, laptops, tablets, personal computers, desktop computers, and smart wearables. Embodiment of the present disclosure are intended to cover or otherwise include various types of electronic devices such as, existing/known, related art, and/or later developed technologies. In a preferred embodiment, the user device 202 may be operationally coupled to a communication network for connecting with the other components of the system 200. The communication network may include various wireless communication technologies and networking technologies including, existing/known, related art, and/or later developed technologies. In a preferred embodiment, the user device 202 via a user device connects to the other components of the system 200 hosted on a remote server via the communication network.
[0075] In an embodiment of the present disclosure, the user intent recognition module 204 may handles incoming user queries and processes them to understand the intent by leveraging various natural language processing (NLP) to interpret the user's request.
[0076] In an embodiment of the present disclosure, the artificial intelligence (AI)-backend infrastructure 208 use a combination of VectorDB and the semantic data cache to quickly access relevant information and generate contextually appropriate replies.
[0077] In an embodiment of the present disclosure, the structured query language (SQL) generator 212 may generates structured query language (SQL) queries to retrieve specific information from the data warehouse 210. In an embodiment of the present disclosure, the dashboarding tools providing visualization and reporting capabilities via the brand owner interface 202 may help in monitoring key performance indicators (KPIs) like, Average Order Value (AOV), Lifetime Value (LTV), and others.
[0078] Fig. 2B illustrates a block diagram of an exemplary system architecture for the system 200 for caching user intents and semantic data for personalized responses in conversational artificial intelligence (AI)-platform, in accordance with an exemplary embodiment of the present disclosure.
[0079] In an exemplary embodiment, the system 200 that may be trademarked with a name “Ulai”, a conversational commerce platform that may integrate various components to enhance customer interaction and support ecommerce functions.
[0080] In an exemplary embodiment, the user may send a query through a messaging platform. The system 200 may capture the query and processes it for intent recognition and various natural language processing models may analyse the query to determine the user's intent. The identified intents are tagged and stored in the semantic data cache 216. The recognized intent may be stored in the semantic data cache 216 along with related semantic data. The system 200 may maps this cached data to user attributes to enhance personalization. The user attributes may include, but not limited to preferences, past behaviours, and demographic information.
[0081] In an exemplary embodiment, whenever a new query is received, the system 200 checks the data semantic cache 216 for pre-stored intents and associated data. If a match is found, the system 200 retrieves the relevant data and generates a personalized response. The response is optimized to reduce token utilization, improving efficiency and response time for the system 200.
[0082] In an embodiment of the present disclosure, the system 200 may perform predictive pre-fetching with the predictive model embedded with the artificial intelligence (AI)-backend infrastructure 208 analyses historical interaction data and user attributes to predict future intents. The relevant responses are pre-fetched and stored in the semantic data cache 216.
[0083] In an exemplary embodiment, the user query may be "Get me the top 5 trending products" and the system 200 may process this query by fetching data based on criteria like, creation date, product sales, and product visits.
[0084] In some embodiments, the user intent recognition module 204 may use a natural language processing engine to derive context or intent from the incoming user data. In some embodiments, the user intent recognition module 204 may have multilingual capabilities to cater to users using different languages. In some embodiments, the natural language processing engine may convert input voice data/query to a textual form before chat processing.
[0085] In some embodiments, the system 200 may include a feedback module allowing the user to rate the responses generated and incorporate the feedback received to improve the overall performance of the system over time. In some embodiments, the structured query language (SQL) generator 212 may optimize query performance based on historical query patterns and data access statistics. In some embodiments, the brand owner interface 214 may supports user role management and allow customization of the reports generated or the dashboard as per the requirement. In some embodiments, the artificial intelligence (AI)-backend infrastructure (208) may be a third-party AI services.
[0086] The proposed invention offers several advantages. The primary advantage of the proposed invention is improved response time and real time personalisation leads to seamless user experience. Efficient resource utilisation that aids in handling handle large number of concurrent users and the use of the semantic data cache 216 helps in maintaining stateful interaction leading to more coherent conversations with user. The response time for the proposed system 200 and method 100 utilizing the semantic data cache 216 is 1.5 to 4 seconds as compared to the traditional art having the response time of 7 to 20 seconds.
[0087] The proposed invention may enhance personalization by analysing responses post-generation, the predictive model embedded with the artificial intelligence (AI)-backend infrastructure 208 can generate better tailor future interactions based on more comprehensive user data and behaviour patterns. The proposed invention may enhance efficiency in pre-fetching by placing the predictive model after initial response generation allows it to use more recent interaction data, making predictions and pre-fetching more accurate.
[0088] The proposed invention may provide improved user experience with the users benefiting from faster and more relevant responses, as the system 200 can quickly adapt to changing user intents and preferences. The method 100 provides a clear and efficient process for generating personalized responses in a conversational AI platform, ensuring that both current and future user interactions are handled optimally.
[0089] Fig. 3 illustrates a flowchart for a hybrid graph-based caching method 300 for optimizing handling of the chatbot queries and responses, in accordance with an exemplary embodiment of the present disclosure. The method 300 may comprise following steps.
[0090] At 302, constructing a dynamic graph structure where a plurality of node represents queries and user profiles, interconnected by a plurality of edge.
[0091] In an embodiment of the present disclosure, the edges may include profile-profile edges having weights based on profile similarity, overlap score, interaction frequency, and graph-based metrics, query-query edges having weights based on semantic similarity and graph-based metrics, and profile-query edges having weights based on relevance score, frequency count, recency score, and contextual relevance.
[0092] At 304, performing cache lookup/search by using a plurality of feature-based processes, employing a plurality of graph-based similarity measures, and thereafter performing cache indexing.
[0093] In an embodiment of the present disclosure, the feature-based processes may include semantic similarity between queries and responses, keyword search, and hybrid search. The graph-based similarity measures may be included such as, neighbourhood search, path-based search, graph similarity search, and community detection algorithms identifies clusters or communities of a plurality of user with similar interests or behaviours, the cache indexing may include maintaining an inverted index of cached queries based on the tokens or features and hashing the queries into buckets based on features.
[0094] At 306, updating the cache by continuously updating the dynamic graph structure followed by edge weight update, and community detection update.
[0095] In an embodiment of the present disclosure, the updating the dynamic graph structure may reflect evolving user interactions and query patterns, followed by adjusting edge weights between nodes based on new data or updated metrics, and periodically recalibrating community structures within the graph to reflect shifts in user behaviours or interests.
[0096] At 308, implementing eviction policies by simultaneously performing time-based expiration, usage-based eviction, and size-based eviction.
[0097] In an embodiment of the present disclosure, the time-based expiration may be performed by setting expiration times for cached entries based on the likelihood of query relevance over time. The usage-based eviction may be performed by removing least frequently accessed queries or responses from the cache to make room for new entries. The size-based eviction may be performed by managing cache size by evicting entries when the cache reaches a predefined capacity threshold. Embodiments of the present disclosure are intended to include or otherwise cover all kinds of eviction policies including, known, related art, and/or later developed technologies.
[0098] Fig. 4 illustrates an exemplary operational flow 400 for the hybrid graph-based caching, in accordance with an exemplary embodiment of the present disclosure. The exemplary operational flow may comprise following steps.
[0099] At 402, generating query/user query by a user.
[0100] At 404, receiving the incoming query.
[0101] At 406, checking for a query similarity.
[0102] At 408, searching for a cache hit/match based on the query similarity.
[0103] At 410, searching for a user profile, if no cache hit/match is found.
[0104] At 412, initiating personalization of response, if cache hit/match is found.
[0105] At 414, generating final response.
[0106] At 416, checking if the user profile exists.
[0107] At 418, generating query similarity based on the user profile, if the user profile exists.
[0108] At 420, searching for a cache hit/match.
[0109] At 422, performing community detection, if cache hit/match is found.
[0110] At 424, personalizing response based on the community.
[0111] At 426, generating final response.
[0112] At 428, sending the user query to a chat engine, if no cache hit/match is found.
[0113] At 430, generating final response through the chat engine.
[0114] At 432, continuously updating the graph.
[0115] In an embodiment of the present disclosure, the implementation of graph structure and the selection of graph database may ensure scalability.
[0116] While the invention has been described in connection with what is presently considered to be the most practical and various embodiments, it will be understood that the invention is not to be limited to the disclosed embodiments, but on the contrary, is intended to cover various modifications and equivalent arrangements included within the scope of the appended claims. A person of ordinary skill in the art may be aware that, in combination with the examples described in the embodiments disclosed in this specification, units and algorithm steps may be implemented by electronic hardware, computer software, or a combination thereof.

, Claims:WE Claim:
1. A method (100) for caching user intents and semantic data for personalized responses in conversational artificial intelligence (AI)-platform, the method (100) comprising:
receiving a user query through a messaging platform hosted on a user device (202),
wherein a plurality of user query is handled simultaneously across a plurality of messaging platforms;
analysing the user query using a plurality of natural language processing (NLP) technique to recognize the user intent;
performing cache lookup,
wherein the cache lookup involves checking a semantic data cache (216) for pre-stored intents and associated data corresponding to the identified user intent;
storing the recognized intent in the semantic data cache (216) along with associated semantic data and user profile information;
mapping the cached data with user attributes such as preferences, behaviour patterns, and demographic information;
retrieving relevant data from the semantic data cache (216) and processing the user queries by fetching data based on pre-determined criteria; and
generating a personalized response using the retrieved data.
2. The method (100) as claimed in claim 1, wherein the method (100) performs token optimization by reducing the number of tokens required for processing, and enhancing efficiency, with the use of pre-cached data.
3. The method (100) as claimed in claim 1, wherein the method (100) predicts future user intents based on historical interaction data.
4. The method (100) as claimed in claim 1, wherein the cache lookup further includes:
finding a relevant cached response based on the user query and attributes, if matching occurs; and
forwarding the query to an artificial intelligence (AI)-backend infrastructure (208) for processing, if no matching occurs.
5. The method (100) as claimed in claim 4, wherein the cache lookup also includes checking for pre-fetched data relevant to the current user query.
6. The method (100) as claimed in claim 1, wherein the method (100) also includes visualization and reporting of key performance indicators (KPIs) through a plurality of dashboarding tool on a brand owner device (214).
7. A system (200) for caching user intents and semantic data for personalized responses in conversational artificial intelligence (AI)-platform, the system (200) comprising:
a plurality of user device (202) configured to allow a plurality of user to send a query through a plurality of messaging platform,
wherein the user device (202) also provide response generated to the user;
a user intent recognition module (204) operationally coupled to the user device (202), the user intent recognition module (204) configured to handle incoming user queries through natural language processing (NLP) to understand user intent,
wherein the user intent recognition module (204) performs chat processing by performing a plurality of data operations including normalisation, context awareness, prompt engineering, and multi label classification;
a user profile database (206) operationally coupled to the user intent recognition module (204), the user profile database (206) configured to store and retrieve data related to the profile of the user;
an artificial intelligence (AI)-backend infrastructure (208) operationally coupled to the user intent recognition module (204), the artificial intelligence (AI)-backend infrastructure (208) configured to leverage large language models (LLMs) and openAI to generate intelligent responses,
wherein the artificial intelligence (AI)-backend infrastructure (208) includes a semantic data cache (216) configured to cache recognized user intents and associated semantic data coupled with the user profile data to facilitate personalisation and faster response,
wherein the semantic data cache (216) facilitates faster response generation by storing frequently accessed information;
a data warehouse (210) operationally coupled to the artificial intelligence (AI)-backend infrastructure (208), the data warehouse (210) configured to aggregate data from the user profile database (206) and other sources to provide a unified view for analysis;
a structured query language (SQL) generator (212) operationally coupled to the artificial intelligence (AI)-backend infrastructure (208), the structured query language (SQL) generator (212) configured to generating structured query language (SQL) queries to retrieve specific information from the data warehouse,
wherein the structured query language (SQL) generator (212) is integrated with the artificial intelligence (AI)-backend infrastructure (208) to enhance query formulation and execution;
a band owner interface (214) operationally coupled to the artificial intelligence (AI)-backend infrastructure (208), the band owner interface (214) configured to:
provide insights and actionable data to improve customer engagement and sales strategies; and
allow brand owners to manage customer interactions and/ analyse performance metrics,
wherein band owner interface (214) also includes a plurality of dashboarding tool for visualization and reporting of key performance indicators (KPIs)
8. The system (200) as claimed in claim 7, wherein the semantic data cache (216) uses a cache hit counter operable to:
update cached data; and
remove outdated or irrelevant cached data.
9. The system (200) as claimed in claim 8, wherein the semantic data cache (216) configured to map the cached data with user specific attributes such as, preferences, behaviour patterns, demographics.
10. A hybrid graph-based caching method (300) for optimizing handling of the chatbot queries and responses, the method (300) comprising:
constructing a dynamic graph structure where a plurality of node represents queries and user profiles, interconnected by a plurality of edge;
performing cache lookup/search by using a plurality of feature-based processes, employing a plurality of graph-based similarity measures, and thereafter performing cache indexing;
updating the cache by continuously updating the dynamic graph structure followed by edge weight update, and community detection update; and
implementing eviction policies by simultaneously performing time-based expiration, usage-based eviction, and size-based eviction.