Abstract: Complex computer system architectures are described for analyzing data elements of a knowledge graph, and predicting new facts from relational learning applied to the knowledge graph. This discovery process includes converting the knowledge graph into a set of candidate embeddings spaces to apply further analysis to rank the set of candidate embeddings spaces, where the top ranked candidate embeddings spaces are further processed to identify the new facts.

Abstract: The systems and methods described herein may generate multi-modal embeddings with sub-symbolic features and symbolic features. The sub-symbolic embeddings may be generated with computer vision processing. The symbolic features may include mathematical representations of image content, which are enriched with information from background knowledge sources. The system may aggregate the sub-symbolic and symbolic features using aggregation techniques such as concatenation, averaging, summing, and/or maxing. The multi-modal embeddings may be included in a multi-modal embedding model and trained via supervised learning. Once the multi-modal embeddings are trained, the system may generate inferences based on linear algebra operations involving the multi-modal embeddings that are relevant to an inference response to the natural language question and input image.

Abstract: A device receives a knowledge graph and an ontology for the knowledge graph, and receives a query for information associated with the knowledge graph. The device generates candidate responses to the query, and assigns scores to the candidate responses based on the knowledge graph. The device identifies a particular candidate response, of the candidate responses, based on the scores for the candidate responses, and determines, based on the knowledge graph, a neighborhood of the particular candidate response. The device generates knowledge graph embeddings for the neighborhood of the particular candidate response, and determines a particular neighborhood, with a smallest loss of quality, based on the knowledge graph embeddings. The device generates a reasoning graph based on the ontology and the particular neighborhood, and generates an explanation of the particular candidate response based on the reasoning graph. The device performs an action based the explanation of the particular candidate response.

Abstract: The systems and methods described herein may generate multi-modal embeddings with sub-symbolic features and symbolic features. The sub-symbolic embeddings may be generated with computer vision processing. The symbolic features may include mathematical representations of image content, which are enriched with information from background knowledge sources. The system may aggregate the sub-symbolic and symbolic features using aggregation techniques such as concatenation, averaging, summing, and/or maxing. The multi-modal embeddings may be included in a multi-modal embedding model and trained via supervised learning. Once the multi-modal embeddings are trained, the system may generate inferences based on linear algebra operations involving the multi-modal embeddings that are relevant to an inference response to the natural language question and input image.

Abstract: Complex computer system architectures are described for analyzing data elements of a knowledge graph, and predicting new surprising or unforeseen facts from relational learning applied to the knowledge graph. This discovery process takes advantage of the knowledge graph structure to improve the computing capabilities of a device executing a discovery calculation by applying both training and inference analysis techniques on the knowledge graph within an embedding space, and generating a scoring strategy for predicting surprising facts that may be discoverable from the knowledge graph.

Abstract: A system adapted to receive a knowledge base, which may include drug data, human biological data, drug-drug interactions, protein-protein interactions, gene expression, protein and drug interaction data, genotypic information for cell lines, drug side effects, and disease classification labels. The system may generate a knowledge graph based on the knowledge base, and convert the knowledge graph into embeddings that include points in a k-dimensional metric space. The system may determine a medical effect weighting based on a drug combination query, and update the embeddings of the drug combination. The system may utilize a pooling method to update predicate embeddings. The system may determine polypharmacy scores for the embeddings, and rank the predicted links between a drug combination and side effects.

Abstract: A system for providing integrated monitoring and communications of diagnostic equipment is disclosed. The system may comprise a data access interface, a processor, and an output interface. The data access interface may receive heterogeneous data from a plurality of machine and sensor equipment associated with performance of a system or product. The data access interface may also to receive a user inquiry pertaining to the system and product. The processor may generate a knowledge graph based on the data associated with the system or product, as well as convert the user inquiry into a knowledge graph query by: extracting entities from the user inquiry; extracting relations from the user inquiry to identify relationships between entities; expanding the user inquiry using the knowledge graph and the entities and relations; and translating the inquiry into knowledge graph triples.

Abstract: A system for providing integrated monitoring and communications of diagnostic equipment is disclosed. The system may comprise a data access interface, a processor, and an output interface. The data access interface may receive heterogeneous data from a plurality of machine and sensor equipment associated with performance of a system or product. The data access interface may also to receive a user inquiry pertaining to the system and product. The processor may generate a knowledge graph based on the data associated with the system or product, as well as convert the user inquiry into a knowledge graph query by: extracting entities from the user inquiry; extracting relations from the user inquiry to identify relationships between entities; expanding the user inquiry using the knowledge graph and the entities and relations; and translating the inquiry into knowledge graph triples.

Abstract: A device receives a knowledge graph and an ontology for the knowledge graph, and receives a query for information associated with the knowledge graph. The device generates candidate responses to the query, and assigns scores to the candidate responses based on the knowledge graph. The device identifies a particular candidate response, of the candidate responses, based on the scores for the candidate responses, and determines, based on the knowledge graph, a neighborhood of the particular candidate response. The device generates knowledge graph embeddings for the neighborhood of the particular candidate response, and determines a particular neighborhood, with a smallest loss of quality, based on the knowledge graph embeddings. The device generates a reasoning graph based on the ontology and the particular neighborhood, and generates an explanation of the particular candidate response based on the reasoning graph. The device performs an action based the explanation of the particular candidate response.

Abstract: A device may receive individual information associated with individual activities of an individual, and may aggregate the individual information, based on a time period, to generate aggregated individual information. The device may identify patterns in the aggregated individual information, and may determine states for the patterns based on state information associated with activities capable of being performed by individuals. The device may generate a sequential knowledge graph based on modifying a knowledge graph with the states and adding a sequence of activities to the knowledge graph, and may determine embeddings for the individual activities based on the sequential knowledge graph.

Abstract: A device receives training data and an ontology for the training data, where the training data includes information associated with a subject of the ontology. The device generates a knowledge graph based on the training data and the ontology, and converts the knowledge graph into knowledge graph embeddings, where the knowledge graph embeddings include points in a k-dimensional metric space. The device receives a new entity that is not present in the knowledge graph embeddings, and generates a new embedding of the new entity. The device adds the new embedding to the knowledge graph embeddings, and utilizes the knowledge graph embeddings, with the new embedding, to perform an action.

Abstract: A method, apparatus and program for completing a knowledge graph from a plurality of predicates and associated entities, the predicates each providing information on a relationship between a pair of entities, the method comprising the steps of: receiving an input comprising the plurality of predicates and associated entities; searching an axiom database and identifying predicates among the plurality of predicates that are equivalent to one another, or inverses of one another; identifying further predicates that are related to one another, using the axiom database and identified predicates; and embedding the identified predicates and associated entities into a vector space to complete the knowledge graph.

Abstract: A Data Constraint Engine (100) for enforcing data constraints in a polyglot data tier (20) having a plurality of database-specific data stores (21, 22, 23) of various types such as an RDBMS (21), a Triplestore (22), and a MongoDB (23). The Data Constraint Engine uses the concept of a unified data model based on “records” in order to allow data constraints to be defined (using so-called “record shapes”) in a store-agnostic way. The Data Constraint Engine includes APIs (130) for processing incoming requests from remote clients (30) relating to data in the polyglot data tier, for example a request to create or update data in a data store. The APIs extract, from such a request, a record corresponding to the data specified in the request and a data source identifier identifying the data store holding the specified data. Then, on the basis of the record extracted by the interface, an appropriate record shape is extracted from a shapes catalogue (110), the record shape determining the structure of the record.

Abstract: A query mediator arranged to query a polyglot data tier of data stores, each data store adopting a data model and the polyglot data tier including at least two different types of data store with differing data models. The query mediator including at least one HTTP API; a catalogue containing metadata for each data store; and a plurality of adapters, one for each data model. The API receives an incoming query from a client, checks the query against the catalogue to identify a correct data store storing the queried data, and routes the query to an adapter for the correct data store. The adapter transforms the query into a format suitable for use with the data model adopted in the correct data store, for execution by the relevant data store. The API returns the query result to the client in response to the incoming query.