International Business Machines Corporation

Chengmin Ding

Elinna Shek

Stanley John Vernier

Renee F Decker

Octavian Florin Filoti

Charles Evan Beller

Allen Ginsberg

Adam D Dumey

Daniel M Jamrog

Christopher Munjal Nolan

Methods, computer program products, and systems are presented. The method computer program products, and systems can include, for instance: examining data of a first user interactive user interfacing application for interactive engagement; and examining data of a second user interactive user interfacing application for interactive engagement.

Adaptive user interfacing

Methods, computer program products, and systems are presented. The method computer program products, and systems can include, for instance: examining programming data defining a first user interactive user interfacing application for interactive engagement with an end user; examining programming data defining a second user interactive user interfacing application for interactive engagement with a user; identifying one or more difference between the second user interactive user interfacing application and the first user interactive user interfacing application using result data resulting from the examining programming data defining a first user interactive user interfacing application and resulting data resulting from the examining programming data defining a second user interactive user interfacing application; and generating update data for updating the second user interactive user interfacing application in dependence on the one or more difference between the second user interactive user interfacing application and the first user interactive user interfacing application.

A question answering (QA) system comprising memory for storing instructions, and a processor configured to execute the instructions to ingest source documents that include structured data and unstructured data to create a knowledge base, wherein the unstructured data includes table data; create table annotations to represent the table data; store the ingested structured data, unstructured data, and the table annotations in the knowledge base; and determine answers to questions using the knowledge base.

Framework for analyzing table data by question answering systems

A system for question answering (QA) documents data ingestion decides to ingest the documents data through a first plurality of sub-pipelines including a first sub-pipeline having a first set of engines and a second sub-pipeline having a second set of engines being independent from the first set of engines. The system determines a subset of the documents data and decides to ingest the subset through a second plurality of sub-pipelines including a third sub-pipeline having a third set of engines and a fourth sub-pipeline having a fourth set of engines being independent from the third set of engines. A set of engines of the second plurality of sub-pipelines and a set of engines of the first plurality of sub-pipelines are in a common class. The system selects output data from the second plurality of sub-pipelines over corresponding output data from the first plurality of sub-pipelines and generates a knowledge base.

Plug-and-ingest framework for question answering systems

An approach is provided in which the approach provides a first question to a requestor requesting access to a resource. The first question corresponds to a first cognitive level and includes at least one image selected from a set of images. The approach computes a confidence value of a first answer received from the requestor responding to the first question. In response to determining that the confidence value is below a confidence threshold, the approach provides a second question to the requestor corresponding to a second cognitive level that is increased from the first cognitive level. The approach grants access to the requestor in response to determining that a second answer received from the requestor responding to the second question is a correct answer.

Detector for online user verification

A method for supplementing context in dialog flows for chatbot systems includes responsive to receiving identification information associated with a user and a topic for a chatbot conversation, initializing a chatbot conversation between a chatbot and the user. The method identifies a set of entities and a set of one or more relationships, wherein each relationship from the set one or more relationships is between two entities from the set of entities. The method determines an initial set of relevancy scores for each entity in the set of entities and the set of one or more relationships, wherein the initial set of relevancy scores are based at least on a domain for the topic of the chatbot conversation. The method generates a first response to the user based on the initial set of relevancy scores, wherein the first response includes at least one entity and at least one relationship.

Context retrieval for chatbots

A system includes a memory having instructions stored therein. The system also includes at least one processor in communication with the memory. The at least one processor is configured to execute the instructions to receive an input question in a source language, generate a first similarity score based at least in part on a source-language operand, generate a target-language operand based at least in part on the source-language operand, generate a second similarity score based at least in part on the target-language operand, generate a semantic concept score based at least in part on the first similarity score and the second similarity score, generate a set of ranked answers to the input question ('ranked answer set') based at least in part on the semantic concept score, and output the ranked answer set.

Semantic concept scorer based on an ensemble of language translation models for question answer system

A system for handling a graphical representation of data associated with a question answering (QA) input document includes a memory having instructions therein and includes at least one processor in communication with the memory. The at least one processor is configured to execute the instructions to derive, at least from a portion of the QA input document, first metadata regarding a context of the graphical representation of data. The at least one processor is also configured to execute the instructions to derive, at least from a portion of the graphical representation of data, tabular data. The at least one processor is also configured to execute the instructions to determine, at least in part by comparing at least a portion of the first metadata to existing table annotations from a QA knowledge base, how to incorporate the tabular data into the QA knowledge base.

Framework for analyzing graphical data by question answering systems

A method trains an artificial intelligence system to respond to a complex query. A query/answer (Q/A) system sends a first query to an information server, where the first query is a query about a first topic with regard to a first entity, and where the information server provides access to unstructured information sources. In response to receiving a negative response to the first query for the first entity, the Q/A system sends variations of the first query and a second query about a second topic to the information server. The Q/A system identifies a pattern in the table of information in order to create the answer to the first query for the first entity. The Q/A system then trains an artificial intelligence system to respond to a complex query about the first entity with information in the table of information.

Training an artificial intelligence to generate an answer to a query based on an answer table pattern

Embodiments can provide a computer implemented method, in a data processing system comprising a processor and a memory comprising instructions which are executed by the processor to cause the processor to implement a media artifact annotation system, the method comprising inputting one or more relationships; for each of the one or more relationships, extracting, through an entity argument extraction module, one or more entity arguments; constructing, through a media query construction module, a media query using the one or more entity arguments; submitting the media query to a media search corpus; receiving search results comprising one or more media artifacts from the media search corpus; passing, through an annotation module, the search results to an annotator; receiving, through a response input module, one or more responses regarding each of the one or more media artifacts from the annotator; if the response for the media artifact is confirmatory, applying an annotation to the media artifact; and if the response for the media artifact is rejection, applying no annotation to the media artifact.

System and method for rapid annotation of media artifacts with relationship-level semantic content

A method of augmenting a knowledge graph includes obtaining the knowledge graph, which includes entities and relationships between the entities defining respective edges, clustering the entities into knowledge domains using semantic distances determined between the entities and a threshold on the semantic distances, identifying strengths of the relationships between adjacent entities in the knowledge graph, creating knowledge chains from node pairs in the knowledge graph, including generating a minimum spanning tree using the strengths of the relationships, pruning edges from the knowledge chain using a threshold on weights corresponding to the edges, defining a first knowledge index for each of the knowledge chains, defining a second knowledge index for each of the knowledge domains, and defining a third knowledge index for the knowledge graph as a harmonic mean of a sum of the first knowledge indexes and a sum of the second knowledge indexes.

Multi-dimensional knowledge index and application thereof

A system, computer program product, and method are provided to automate a natural language processing system to facilitate an artificial intelligence platform defining a relationship between dialogue and post dialogue activity. Dialogue is detected and analyzed, including identification of key words and phrases within the dialogue. Post dialogue actions, including physical actuation of a hardware device and an associated temporal proximity of the action and the dialogue, are monitored. The hardware device receives an instruction from a processing unit that relates to the analyzed dialogue and the hardware device changes states and/or actuates another hardware device. The system constructs a hypothesis, i.e., a relationship from the identified key phrase drawn from the analyzed dialogue and the monitored post action dialogue. A dialogue tree containing identified terms and associated post dialogue actions is dynamically modified with one or more new identified terms and the associated post dialogue actions.

Leveraging natural language processing

A technique for responding to user input includes assigning respective initial confidence scores to relationship n-tuples in a knowledge graph (KG). Each of the n-tuples designates at least a first entity, a second entity, and a relationship between the first and second entities or a single entity and a relationship between the single entity and one or more properties of the single entity. Respective feature vectors are associated with each of the n-tuples. A training set that includes at least a subset of the n-tuples labeled with respective ground truth labels is generated. Respective initial weights are learned for the feature vectors based on the training set. Respective subsequent confidence scores are generated for each of the n-tuples based on the initial weights for the feature vectors. A response to user input is generated based on the subsequent confidence scores for one or more of the n-tuples.

Responding to user input based on confidence scores assigned to relationship entries in a knowledge graph

A technique for assigning confidence scores to relationship entries in a knowledge graph includes assigning respective initial confidence scores to relationship n-tuples in a knowledge graph. Each of the relationship n-tuples designates at least a first entity, a second entity, and a relationship between the first and second entities or a single entity and a relationship between the single entity and one or more properties of the single entity. Respective feature vectors are associated with each of the relationship n-tuples. A training set that includes at least a subset of the relationship n-tuples labeled with respective ground truth labels is generated. Respective initial weights are learned for the feature vectors based on the training set. Respective subsequent confidence scores are generated for each of the relationship n-tuples based on the initial weights for the feature vectors.

Techniques for assigning confidence scores to relationship entries in a knowledge graph

Natural language processing of raw text data for optimal sentence boundary placement. Raw text is extracted from a document and subject to cleaning. The extracted raw text is examined to identify preliminary sentence boundaries, which are used to identify potential sentences in the raw text. One or more potential sentences are assigned a well-formedness score. A value of the score correlates to whether the potential sentence is a truncated/ill-formed sentence or a well-formed sentence. One or more preliminary sentence boundaries are optimized depending on the value of the score of the potential sentence(s). Accordingly, the processing herein is an optimization that creates a sentence boundary optimized output.

Growing community of inventors

Chantilly, VA, United States of America

Chengmin Ding