Systems and methods of evaluating and fine-tuning a generative AI tool on a communication platform. The communication platform accesses a dataset comprising a user query and a response generated by an AI-based query system. The communication platform evaluates the response with respect to the user query using multiple AI-based scoring models to obtain multiple evaluation results. In response to determining that the multiple evaluation results are inconsistent, the communication platform evaluates the response with respect to the user query using a reference large language model (LLM) to provide a reference evaluation result. In response to determining that the reference evaluation result is decisive, the communication platform classifies, based on the reference evaluation result, the dataset to a data category of one or more data categories. The communication platform fine-tunes the AI-based query system based on a group of datasets in the data category.

A system includes a storage device and at least one processor in communication with the storage device. The at least one processor receives a query associated with a plurality of data tables stored in the storage device. The at least one processor processes the query using a large language model (LLM) trained on semantic mapping information that describes relationships between data elements stored within the plurality of tables. The at least one processor generates, with the LLM, a natural language response to the query based on semantic mapping data generated from the data elements stored withing the tables. A method and computer-readable medium are also disclosed.

System and method for managing information compliance and relevance using autonomous artificial intelligence (AI) agents in data transfer and communication environments. Some embodiments may include a core orchestration engine with multiple autonomous AI agents configured to manage and evaluate the compliance and relevance of information in communication and data transfer environments. The system may use weighted metrics to assess if information and actions comply with regulations and are pertinent to recipients, monitor email and data transfer, ensure regulatory compliance, and enhance information and knowledge sharing within organizations. The system may use semantic embeddings, part-of-speech analysis, and language models to extract and apply regulatory rules efficiently. These features may significantly reduce search space, computational overhead, and manual effort while improving security and accuracy.

The disclosure generally describes methods, software, and systems for to data access control for applications using large language models (LLM). A request to access attributes of an application object is received from a user. A LLM object access layer defining an access level of the user to each attribute of the application object is determined. The LLM object access layer is queried to determine accessible attributes of the attributes of the application object. A response is provided based on the accessible attributes of the application object to the user.

Systems, articles, and computer-implemented methods are disclosed for generating natural language summaries of a multi-dimensional analysis of a detected anomaly within a member of multi-dimensional data by prompting a LLM with a prompt generated to include data about the anomaly in a manner understandable by the LLM. The prompt to the LLM includes a path to a member of the hierarchy containing an anomaly with a delimiter between the member and ancestor nodes. The delimiter allows the ancestral context of the member of the hierarchy to be understood by the LLM. The prompt also includes a metric defining a magnitude of the anomaly in relation to another value, such as an average, a value of the anomaly, a time corresponding to the anomaly, and one or more examples of other anomalies with included data about those anomalies matching the type of data provided for the detected anomaly.

The present disclosure relates to a knowledge question and answer method, a computer-readable medium and an electronic device, the method includes: acquiring a target question input by a user in a natural language; retrieving, by a machine learning model, in a first knowledge base according to the target question to obtain a target data field for answering the target question, and retrieving in a second knowledge base according to the target question to obtain a target knowledge document for answering the target question, and determining a target answer to the target question according to the target data field and the target knowledge document, where the first knowledge base is configured to store metadata fields of a business data table composed of business data, and the second knowledge base is configured to store business knowledge documents; and displaying the target answer to the user.

The present disclosure relates to a data processing method, an apparatus, a medium, a device and a computer program product. The method includes: receiving retrieval data; matching the retrieval data with structured data in a knowledge database to obtain candidate question data and candidate answer data, where knowledge in the knowledge database is represented by the structured data, and each piece of the structured data includes one piece of answer data and at least one piece of question data corresponding to the answer data; and generating a target retrieval result corresponding to the retrieval data according to the candidate question data and the candidate answer data.

A product search method includes: providing a product search interaction interface including a first area and a second area, the first area being configured to provide an Artificial Intelligence (AI) interaction component for: receiving a user's search request expressed through inputting a natural language statement; conducting multi-round interactions with a large AI model, including receiving refined expression statements of the user's search request inspired by the large AI model's responses to clarify the user's search request, and the second area for: providing attribute options for clarifying the search request based on interactions in the first area and corresponding attribute value options under each attribute option; and receiving user's search request information expressed by selecting one or more attributes and attribute values. The input statements collected from the first area and the attribute and attribute value selection results from the second area are fused to provide the product search result.

In general, according to one embodiment, an interactive data processing apparatus includes a processor including hardware. The processor receives a user input. The processor determines an abstraction level of the user input. The processor generates a question for a user to reduce the abstraction level in a case where the abstraction level is determined to be high. The processor determines data processing necessary for outputting a result according to the user input in a case where the abstraction level is determined to be low.

Systems, articles, and computer-implemented methods are provided for generating summaries of a plurality of insights in multi-dimensional data to describe underlying trends using a large language model. A data structure is generated describing the plurality of insights where the data structure encapsulates for each insight of the plurality of insights to be included: a member of a data hierarchy that fits a descendant dimension that includes the insight, a value of the descendant dimension that fits the insight, and a characteristic of the insight. The data structure is included within a prompt to a large language model to summarize the plurality of insights. The prompt may also include data representing a relationship between the plurality of insights, such as how a first insight of the plurality of insights contributes to a second insight of the plurality of insights.