Abstract: Aspects of the disclosure are directed to a foundational model for learned database tasks. The foundational model can be task-independent and used with low overhead to accomplish various learned database tasks. The foundational model can include a plurality of expert sub-models, each pre-trained to be utilized alone or in combination based on the database task. The foundational model can accomplish the various learned database tasks with reduced training overhead by relying on one or more particular expert sub-models based on the database task at hand.

Abstract: An offline adaptation method of customizing and incorporating external knowledge domain specific knowledge sources into an existing knowledge base (KB) of a cognitive conversation system, includes obtaining an instance-concept frequency from the existing KB. All the contexts of a domain ontology in which an instance-concept is used are computed. All instance-concepts from the existing KB to an external domain-specific knowledge source are mapped. The existing KB is configured to search for semantically related terms in response to a natural language query received from a conversation system.

Abstract: A computing device and computer-implemented method to provide access to business intelligence data via a conversational interface is described. The method begins with receiving prior user interactions based on corresponding pairs of user queries and computing device responses, with the conversational interface. Received also is a business intelligence ontology and access to a conversational space structure for the conversational interface. The computing device executes a machine learning model based upon the prior user interactions, business intelligence ontology, and the conversational space structure to predict possible next queries. The possible next queries are sent to a user. In one example, the computing device receives a user query, which is in addition to the prior user interactions with the conversational interface. A response to the query is generated. The response to the query is sent to the user prior to sending the possible next queries, which have been predicted.

Abstract: Techniques for distributed data placement are provided. Query workload information corresponding to a domain is determined by a data orchestrator, and the query workload information is modeled as a hypergraph, where the hypergraph includes a set of vertices and a set of hyperedges, where each vertex in the set of vertices corresponds to a concept in an ontology associated with the domain. Mappings are generated between concepts and a plurality of data nodes based on the hypergraph and based further on predefined capability of each of the plurality of data nodes. A distributed knowledge base is established based on the generated mappings.

Abstract: Aspects of the disclosure are directed to a foundational model for learned database tasks. The foundational model can be task-independent and used with low overhead to accomplish various learned database tasks. The foundational model can include a plurality of expert sub-models, each pre-trained to be utilized alone or in combination based on the database task. The foundational model can accomplish the various learned database tasks with reduced training overhead by relying on one or more particular expert sub-models based on the database task at hand.

Abstract: An embodiment includes extracting, responsive to an update request from a remote requesting system, technical descriptor data from a data source. The embodiment also includes forming a new graph data structure using the technical descriptor data extracted from the data source. The embodiment also includes augmenting the new graph data structure to include a concept based on a value from instance data from the data source. The embodiment also includes identifying a first pair of concepts that are connected in a pre-existing ontology that correspond with a second pair of concepts that lack a connection therebetween in the new graph structure. The embodiment also includes augmenting the new graph data structure to include a connection between the second pair of concepts. The embodiment also includes outputting the new graph data structure as part of a response to the update request from the requesting system.

Abstract: Computer-implemented techniques for entity disambiguation using graph neural networks (GNNs) are provided. According to an embodiment, computer implemented method can comprise receiving, by a system operatively coupled to a processor, an unstructured text snippet comprising an unknown term. The method further comprises employing, by the system, a heterogeneous GNN trained on a knowledge graph associated with a domain of the unstructured text snippet to facilitate identifying one or more similar terms included within the knowledge graph for the unknown term.

Abstract: An insight comprising a fact about a portion of data in a dataset is detected in the dataset. A context is determined by analyzing a natural language input, the context identifying an intent and an entity referenced by the natural language input. A relevance score is computed for the insight, the relevance score comprising a measure of similarity between a vector representation of the insight and a vector representation of the context, the vector representation of the insight comprising an encoding of the insight in multidimensional numerical form, the vector representation of the context comprising an encoding of the context in multidimensional numerical form. A selected insight is presented, the selected insight selected from a set of insights including the insight, the selected insight selected based on the relevance score.

Abstract: A computer-implemented method for generating an ontology-driven conversational interface includes generating an ontology from a description of a domain schema of a Data Analysis (DA) model, in which the DA model is a defined in terms of quantifiable, qualifying or categorical entities and their relationships as described by the domain schema. Conversational artifacts of a conversation space including a conversational pattern framework are generated by extracting DA-related intents, entities, and a dialog from the generated ontology for the conversational interface. A dialog logic table maps DA-related patterns to intents, extracted quantifiable, qualifying or categorical attributes to entities, and the dialog to user-prompts for one or more parameters in an identified DA pattern. The conversation space is integrated with at least one of an external data source or an analytics platform that stores and processes data.

Abstract: An offline adaptation method of customizing and incorporating external knowledge domain specific knowledge sources into an existing knowledge base (KB) of a cognitive conversation system, includes obtaining an instance-concept frequency from the existing KB. All the contexts of a domain ontology in which an instance-concept is used are computed. All instance-concepts from the existing KB to an external domain-specific knowledge source are mapped. The existing KB is configured to search for semantically related terms in response to a natural language query received from a conversation system.

Abstract: A system, method, and computer readable medium perform a method of query relaxation. A query output from a conversational system is received. At least one search term in the query is identified. Instance data is output to the conversational system in response to determining that the instance data in a data store matches the at least one search term in the query. The computer device outputs the received query to an external domain-specific knowledge source in response to determining that the at least one search term does not the match instance data in the data store. The computer device receives the relaxed data matches from the external domain-specific knowledge source being semantically-related to at least one search term in the query based on a plurality of criteria associated with the query. The computing device generates a response to the query based on contextual information and structural information.

Abstract: A computing device and computer-implemented method to provide access to business intelligence data via a conversational interface is described. The method begins with receiving prior user interactions based on corresponding pairs of user queries and computing device responses, with the conversational interface. Received also is a business intelligence ontology and access to a conversational space structure for the conversational interface. The computing device executes a machine learning model based upon the prior user interactions, business intelligence ontology, and the conversational space structure to predict possible next queries. The possible next queries are sent to a user. In one example, the computing device receives a user query, which is in addition to the prior user interactions with the conversational interface. A response to the query is generated. The response to the query is sent to the user prior to sending the possible next queries, which have been predicted.

Abstract: An insight comprising a fact about a portion of data in a dataset is detected in the dataset. A context is determined by analyzing a natural language input, the context identifying an intent and an entity referenced by the natural language input. A relevance score is computed for the insight, the relevance score comprising a measure of similarity between a vector representation of the insight and a vector representation of the context, the vector representation of the insight comprising an encoding of the insight in multidimensional numerical form, the vector representation of the context comprising an encoding of the context in multidimensional numerical form. A selected insight is presented, the selected insight selected from a set of insights including the insight, the selected insight selected based on the relevance score.

Abstract: An embodiment includes extracting, responsive to an update request from a remote requesting system, technical descriptor data from a data source. The embodiment also includes forming a new graph data structure using the technical descriptor data extracted from the data source. The embodiment also includes augmenting the new graph data structure to include a concept based on a value from instance data from the data source. The embodiment also includes identifying a first pair of concepts that are connected in a pre-existing ontology that correspond with a second pair of concepts that lack a connection therebetween in the new graph structure. The embodiment also includes augmenting the new graph data structure to include a connection between the second pair of concepts. The embodiment also includes outputting the new graph data structure as part of a response to the update request from the requesting system.

Abstract: An embodiment includes generating a first concept representation based on a first portion of a knowledge base using a first processing path of a neural network that includes a hyperbolic graph convolution layer. The embodiment also includes generating a second concept representation based on a second portion of the knowledge base using a second processing path of the neural network, the second processing path comprising a heterogenous graph convolution layer. The embodiment also includes generating a unified concept representation including concatenating the first concept representation with the second concept representation. The embodiment also includes generating a prediction score using a using a predictive matching module, where the predictive score is indicative of an extent of a match between the unified concept representation and a concept representation from a second knowledge base.

Abstract: Computer-implemented techniques for entity disambiguation using graph neural networks (GNNs) are provided. According to an embodiment, computer implemented method can comprise receiving, by a system operatively coupled to a processor, an unstructured text snippet comprising an unknown term. The method further comprises employing, by the system, a heterogeneous GNN trained on a knowledge graph associated with a domain of the unstructured text snippet to facilitate identifying one or more similar terms included within the knowledge graph for the unknown term.

Abstract: A method is provided to implement a conversational system with artifact generation. A middleware component receives a user input and determines whether there is sufficient information in the user input and a conversation space in a context storage of the conversational system to identify user intent associated with the user input. Responsive to the middleware component determining there is not sufficient information to identify user intent, a communications handler component sends a natural language query to an external data source via a natural language query (NLQ) interface and receives a natural language response from the external data source. The middleware component updates the conversation space based on the natural language response and returns a user response based on the natural language response.

Abstract: A computer-implemented method for generating an ontology-driven conversational interface includes generating an ontology from a description of a domain schema of a Data Analysis (DA) model, in which the DA model is a defined in terms of quantifiable, qualifying or categorical entities and their relationships as described by the domain schema. Conversational artifacts of a conversation space including a conversational pattern framework are generated by extracting DA-related intents, entities, and a dialog from the generated ontology for the conversational interface. A dialog logic table maps DA-related patterns to intents, extracted quantifiable, qualifying or categorical attributes to entities, and the dialog to user-prompts for one or more parameters in an identified DA pattern. The conversation space is integrated with at least one of an external data source or an analytics platform that stores and processes data.

Abstract: According to one embodiment, a computer program product includes a computer readable storage medium having program instructions embodied therewith. The program instructions are executable by a processor to cause the processor to: identify key concepts in a domain ontology, and define a workload as a set of queries. Terms are extracted from each of the queries, and frequent patterns are detected across the set of queries. Maximal patterns are also identified from the frequent patterns detected, while the key concepts and the maximal patterns are used to create a rich entity. Moreover, identifying the key concepts includes performing centrality analysis of concepts extracted from the domain ontology. Other systems, methods, and computer program products are described in additional embodiments.

Abstract: According to one or more embodiments of the present invention, a computer-implemented method for providing a query response includes receiving, by a computing device, a domain-specific knowledge graph. The method further includes generating a first property graph schema, a property graph schema includes vertices, edges, and properties of the domain-specific knowledge graph, wherein the first property graph schema is generated based on an ontology of the domain-specific knowledge graph. The method further includes generating a second property graph schema from a copy of the first property graph schema that is optimized by applying one or more types of relationships in the first property graph schema. The method further includes instantiating a property graph using the second property graph schema. The method further includes receiving a query to obtain particular data from the domain-specific knowledge graph. The method further includes responding to the query using the property graph.