International Business Machines Corporation

Michael Drzewucki

Charles Evan Beller

Christopher Phipps

Edward Graham Katz

Christopher F Ackermann

Kristen Maria Summers

Paul J Chase, Jr

Richard L Darden

Julie T Yu

Andrew Doyle

James M Dixius

David Kenneth Sagal, Ii

Renee F Decker

James E Ramirez

Elinna Shek

The method provides for receiving a plurality of documents including mentions of a target entity from a search query about the entity. The mentions of the target entity are identified in respective documents of the plurality of documents. Content surrounding the one or more mentions of the target entity are extracted with the mentions within the respective documents and form section. A respective document includes a plurality of sections. Metrics of relevance and irrelevance to the target entity are determined within the plurality of sections of the respective documents. A density score is generated for the plurality of sections of the respective documents. A relevancy score is assigned to respective documents of the plurality of documents, based on the density scores of the sections of the respective documents. The documents are ranked based on the relevancy score and presented in an order based on the ranking.

Ranking entity search results based on information density

A computer device receives a request to search a corpus of documents for an entity, wherein the request includes a non-name identifier of the entity. The computing device identifies entries of text within the corpus of documents that reference the non-name identifier. The computing device applies natural language processing (NLP) to content associated with the identified entries within the corpus of documents, wherein the NLP identifies candidate entities associated with the non-name identifier. The computing device selects an entity from the candidate entities based, at least in part, on distances between the candidate entities and references to the non-name identifier in the identified entries. The computing device returns the selected entity to a submitter of the request.

Entity search system

In an approach to generating a digital sentiment signature to characterize an end to a communication, one or more computer processors detect a start of a communication between at least two participants. A computer starts a digital timer of the communication. A computer identifies one or more digital marks of the communication, where the one or more digital marks are a reflection of a sentiment of at least one of the at least two participants in the communication. A computer generates a digital sentiment signature based on the digital timer and on the one or more digital marks, where the digital sentiment signature is a digital signal that can be communicated across a plurality of types of communication channels. A computer detects an end of the communication. A computer determines a reason for the end of the communication. A computer stores the reason.

Digital sentiment signature generation

Altering messaging according to sentiment analysis by providing a baseline messaging tree including topics arranged according to a sentiment score, providing speaker prompts according to the baseline messaging tree, receiving real-time audience sentiment feedback during a messaging presentation, and adjusting the speaker prompts according to the baseline messaging tree and the real-time audience sentiment feedback.

Altering messaging using sentiment analysis

Approaches presented herein enable presenting, as an initial response to a search query from a user, a set of analytic summaries prior to initiating a document security protocol in a secured document environment. Each analytic summary represents a set or cluster of documents. A summary relevance value is calculated for each analytic summary and a document relevance value is calculated for each individual document. Analytic summaries are presented to the user while each document within each analytic summary are prioritized for a security review based on its related summary relevance value and document relevance value. Based on the priorities, a security review is initiated for the documents so that a highly relevant document within an analytic summary undergoes the security review before a less relevant document. Access to a particular document is granted to the user according to the security review of the document.

Tiered retrieval of secured documents

An embodiment for extracting information from semi-structured text is provided. The embodiment may include identifying one or more high confidence alignments of one or more entities and identifiers in a set of documents. The embodiment may also include analyzing one or more blocks of semi-structured text containing the one or more entities and identifiers. The embodiment may further include identifying one or more known alignments in each of the one or more blocks of semi-structured text. The embodiment may also include generating a structure template. The embodiment may further include applying the structure template to each of the one or more blocks of semi-structured text. The embodiment may also include annotating the set of documents with metadata reflecting the structure template and a location of each of the one or more blocks of semi-structured text.

Flexible pseudo-parsing of dense semi-structured text

Embodiments for managing chatbots are provided. A set of documents is received. A plurality of entities are identified within the set of documents. At least one of the plurality of entities is selected based on a rareness criteria. Contextual data associated with each of the selected at least one of the plurality of entities is identified within the set of documents. At least one question-answer (QA) pair associated with each of the selected at least one of the plurality of entities is generated based on the identified contextual data.

Methods and systems for managing chatbots with respect to rare entities

A method, computer system, and a computer program product for converting a plurality of factoid answers into a plurality of structured relations for storage in a structured knowledge base is provided. The present invention may include receiving a query from a user. The present invention may also include generating a plurality of possible factoid answers. The present invention may then include determining a plurality of confidence scores for the plurality of possible factoid answers. The present invention may then include generating a plurality of certain factoid answers from the plurality of possible factoid answers. The present invention may then include identifying a plurality of special target items. The present invention may further include generating a plurality of structured relations for each certain factoid answer and the identified plurality of special target items. The present include may also include storing the generated plurality of structured relations into the structured knowledge base.

Post-processor for factoid answer conversions into structured relations in a knowledge base

Embodiments for generating a timeline associated with an entity are provided. A plurality of events referenced in a corpus are identified. Each of the plurality of events is associated with an entity. An event time is determined for each of the plurality of events based on information associated with at least one reference to the respective event within the corpus. A score is calculated for each of the plurality of events based on at least one of a number of references to the respective event within the corpus and information associated with a user. A timeline for at least some of the plurality of events is generated based on the calculated scores of the plurality of events and the determined event time for each of the plurality of events.

Methods and systems for generating timelines for entities

Query service receives a query comprising at least a name component. The query service searches a document corpus to identify multiple passages, each comprising a mention of the name component within a selection of one or more documents of the document corpus. The query service collects bins, each bin comprising a distinct selection of the passages from the one or more documents, each of the bins identifying a separate relationship the name component participates in within the distinct selection of passages. The query service assesses a separate score of each respective bin reflecting the relevance of each respective bin to the query. The query service returns a response to the query with the bins each ranked according to each separate score.

Ranking collections of document passages associated with an entity name by relevance to a query

Identifying alternative names of entities includes receiving a query associated with a name of an entity and a corpus. References to the name of the entity within the corpus are identified. The references to the name of the entity within the corpus are analyzed to determine at least one first alternative name of the entity. References to the determined at least one first alternative name of the entity within the corpus are identified. The references to the determined at least one first alternative name of the entity within the corpus are analyzed to determine at least one second alternative name of the entity.

Methods and systems for iterative alias extraction

A method, system and computer program product are provided for implementing an interface for rapid ground truth binning. A set of documents are received wherein each document has at least one entity in a set of entities. A user interface is provided for each received document allowing a user to view passages and select options related to confirming or denying an equivalence between the entity in the received document and an output document entity bin including the entity. Responsive to the user utilizing the user interface and confirming the equivalence, combining the received document with the output document entity bin with reference to the entity.

Implementing interface for rapid ground truth binning

A method includes searching an initial set of documents for an initial set of query names. Each query name of the initial set of query names is associated with at least one document of the initial set of documents. The method also includes prioritizing the initial set of query names based on at least one topic label. The method also includes searching an additional set of documents to generate candidate query names. The method also includes prioritizing the candidate query names based on the at least one topic label. The method further includes applying a temporal search filter to each candidate query name to determine whether the candidate query name was processed within a time frame. The method further includes performing disambiguation processing on each candidate query name not processed within the time frame.

Candidate name generation

A query system identifies a collection of discovered entity bins each comprising unstructured documents with mentions of a name element from a name query and each identified with a particular named entity identifiable from the name element. The query system identifies, from a knowledge base of structured documents, based on identifier components with the name element, candidate records identifying the respective identifier components with the name element, the one or more identifier components identified among the discovery entity bins. For each respective selection of candidate records associated with each bin, the query system applies one or more alignment threshold rules to rank the likelihood that each candidate record within each respective selection matches one or more characteristics of the respective discovery entity bin. The query system aligns, with each of the discovery entity bins, a highest ranked record from among each respective selection of candidate records, where the respective aligned highest ranked record identifies a distinct named entity from among the named entities.

Query-directed discovery and alignment of collections of document passages for improving named entity disambiguation precision

An apparatus comprising a memory and a processor configured for semi-autonomous natural language processing domain adaptation related activities. The processor coupled to the memory and configured to identify a corpus of documents of an evaluation domain and generate a first lexicon based on the corpus of documents of the evaluation domain, and determine a threshold that indicates a sufficiency of domain adaptation of the evaluation domain based at least in part on the first lexicon. The processor is further configured to identify a corpus of documents of a client domain, generate a second lexicon based on the corpus of documents of the client domain, determine a metric associated with the corpus of documents of the client domain and the second lexicon, and determine that domain adaptation of the client domain is complete when the metric exceeds the threshold.

Growing community of inventors

Woodbridge, VA, United States of America

Michael Drzewucki