International Business Machines Corporation

Huaiyu Zhu

Yunyao Li

Sriram Raghavan

Shivakumar Vaithyanathan

Rajasekar Krishnamurthy

Laura Chiticariu

Marina Danilevsky Hailpern

Adriano Crestani Campos

Rahul Gupta

Krishna Prasad Chitrapura

Jennifer Lee Beckmann

Siddhartha Brahma

Alan Akbik

Ishan Jindal

Ching-Tien Ho

Malous Melissa Kossarian

Prithviraj Sen

Mauricio Antonio Hernandez-Sherrington

Douglas A Smith

Jayram S Thathachar

Frederick Ralph Reiss

Vitaly Feldman

Sebastian Johannes Blohm

William S Ko

Sanjana Sahayaraj

Chenguang Wang

Sudeepa Roy

Ashutosh Garg

Honglei Zhuang

Michael James Wehar

Vivian Yaw-Wen Chu

Prasad M Deshpande

Zihui Li

Systems and techniques that facilitate parameterized neighborhood memory adaptation for semantic role labeling are provided. In various embodiments, a system can comprise a receiver component that can access a semantic role labeling model trained on a training dataset. In various aspects, the system can further comprise an execution component that can retrain a labeler of the semantic role labeling model based on a set of neighborhood parameters learned from the training dataset. In various instances, the execution component can execute, after retraining, the semantic role labeling model on an inputted sentence.

Parameterized neighborhood memory adaptation

One or more systems, devices, computer program products and/or computer-implemented methods of use provided herein relate to regularizing semantic similarity relationships relative to a pair of languages. A system can comprise a memory that stores computer executable components, and a processor that executes the computer executable components stored in the memory, wherein the computer executable components comprise a computation component that generates a transformation comprising a semantic similarity relationship between detected semantic labels of a pair of languages.

Determining semantic relationships of argument labels

Providing a predictive model for a target language by determining an instance weight for a labeled source language textual unit according to a set of unlabeled target language textual units, scaling, by the one or more computer processors, an error between a predicted label for the source language textual unit and a ground-truth label for the source language textual unit according to the instance weight, updating, by the one or more computer processors, network parameters of a predictive neural network model for the target language according to the error, and providing, by the one or more computer processors, the predictive neural network model for the target language to a user.

Zero-shot cross-lingual transfer learning

A computer-implemented method according to one embodiment includes identifying features of a plurality of data instances within a target domain; assigning weights to the plurality of data instances within the target domain, based on similarities among the features; selecting a subset of the plurality of data instances within the target domain, based on the weights; associating expert annotations with respective ones of data instances within the subset; and training a machine learning algorithm, utilizing the subset of the plurality of data instances and associated expert annotations.

Implementing a domain adaptive semantic role labeler

Methods, systems, and computer program products for constraint tracking and inference generation are provided herein. A computer-implemented method includes parsing descriptions of one or more user-provided constraints pertaining to data within a target system, parsing truth value assignments to the user-provided constraints, and deriving a truth value for at least one of the user-provided constraints that does not correspond to a known truth value, wherein said deriving comprises performing a logical inference utilizing known truth values of one or more of the user-provided constraints. The computer-implemented method also includes storing, in a database, (i) the user-provided constraints, (ii) the known truth values, and (iii) the at least one derived truth value, and outputting the at least one derived truth value, one or more identified contradictions among the known truth values, and/or an indication that one or more unknown truth values corresponding to the user-provided constraints remain unknown.

Constraint tracking and inference generation

A training query generation system is usable to generate fully formed training questions from prior search queries, some of which may be fully formed search queries and some of which are not fully formed. The system may identify fully formed questions from a query database stored on a storage device. The query database includes a plurality of search query character string, The system further identifies partially formed questions from the query database, creates question templates from the identified fully formed questions, and stores the question templates in a template database. The system then identifies entities with the partially formed questions, classifies each entity, and stores the classified entities in an entity database. Fully formed questions can then be generated using the question templates from the template database and classified entities from the entity database.

Natural question generation from query data using natural language processing system

One embodiment provides method for constructing a cross-lingual information extraction program, the method including: utilizing at least one processor to execute computer code that performs the steps of: constructing a plurality of language-specific representations from text expressed in a plurality of languages by parsing the text of each language using a language-specific semantic parser; mapping the plurality of language-specific representations to a single cross-lingual semantic representation, wherein the cross-lingual semantic representation encompasses the plurality of languages; and constructing the cross-lingual information extraction program based on the cross-lingual semantic representation. Other aspects are described and claimed.

Cross-lingual information extraction program

One embodiment provides a method for generating a natural language resource using a parallel corpus, the method including: utilizing at least one processor to execute computer code that performs the steps of: receiving, from a parallel corpus, natural language text in a source language and a corresponding translation of the natural language text in a target language, wherein the natural language text in the source language comprises linguistic annotations; projecting the linguistic annotations from the source language natural language text to the target language natural language text; applying one or more filters to remove at least one projected linguistic annotation from the target language natural language text that results in at least one error; selecting at least one target language natural language text having substantially complete linguistic annotations; training a machine learning model using the selected at least one target language natural language text and annotations; and adding, using the trained machine learning model, linguistic annotations to at least one target language natural language text having incomplete linguistic annotations. Other aspects are described and claimed.

Generation of a natural language resource using a parallel corpus

Search results are provided for a search query. A search query including an entity and entity type is received. The entity is parsed into semantic components. Variants are generated based on input from auxiliary information and user configuration information for each of the semantic components. The entity is recomposed in different morphological forms from different variants of the semantic components. At least one morphological form for the entity is presented as a search result.

Providing search query results based on entity variant generation and normalization

Determining variants of a text entity comprises parsing the text entity into semantic components and generating variants for each of the semantic components. The entity is recomposed in different morphological forms from the different variants of the semantic components.

Entity variant generation and normalization

A method for refining a dictionary for information extraction, the operations including: inputting a set of extracted results from execution of an extractor comprising the dictionary on a collection of text, wherein the extracted results are labeled as correct results or incorrect results; processing the extracted results using an algorithm configured to set a score of the extractor above a score threshold, wherein the score threshold balances a precision and a recall of the extractor; and outputting a set of candidate dictionary entries corresponding to a full set of dictionary entries, wherein the candidate dictionary entries are candidates to be removed from the dictionary based on the extracted results.

Refining a dictionary for information extraction

Described herein are methods, systems, apparatuses and products for automatically discovering patterns in a text corpus. An aspect provides extracting at least one context string related to at least one annotator from the at least one text corpus; analyzing the at least one context string for at least one sequence, the at least one sequence comprised of at least one subsequence; determining at least one sequence signature for each at least one sequence by applying applicable rules to the at least one sequence; and grouping the at least one sequence signature into at least one group.

Systems and methods for information extraction using contextual pattern discovery

A text annotation structured storage system stores text annotations with associated type information in a structured data store. The present system persists or stores annotations in a structured data store in an indexable and queryable format. Exemplary structured data stores comprise XML databases and relational databases. The system exploits type information in a type system to develop corresponding schemas in a structured data model. The system comprises techniques for mapping annotations to an XML data model and a relational data model. The system captures various features of the type system, such as complex types and inheritance, in the schema for the persistent store. In particular, the repository provides support for path navigation over the hierarchical type system starting at any type.

System and method for storing text annotations with associated type information in a structured data store

The present invention relates to a methodology to translate exact interpretations of keyword queries into meaningful and grammatically correct plain-language queries in order to convey the meaning of these interpretations to the initiator of the search. The method includes the steps of generating at least one grammatically valid plain-language sentence interpretation for a keyword query from a generated sentence plain-language sentence clauses, wherein the grammatically valid plain-language sentence is based upon differing matching elements, and presenting at least one grammatically valid plain-language sentence interpretation for the keyword query to a keyword query system user for the user's review.

Growing community of inventors

Union City, CA, United States of America

Huaiyu Zhu