Method for Near-Real-Time Prediction, Classification, and Notification of Events in Natural Language Systems

Patent metadata

Field	Value
Publication number	US-2016170997-A1
Application number	US-201615052121-A
Country	US
Kind code	A1
Filing date	Feb 24, 2016
Priority date	Nov 6, 2014
Publication date	Jun 16, 2016
Grant date	—

Abstract

An approach is provided for automatically predicting an event occurrence based on a question from an end user presented using a near-real-time natural language processing (NLP) analysis to generate, score and rank a plurality of event occurrences based on a plurality of question context parameters extracted from the question, one or more user profile parameters for the end user, and the one or more historical questions, answers, and events having a specified spatial and/or temporal proximity to the question which are identified by an information handling system. In the approach, performed by an information handling system, a top ranked event occurrence from the ranked plurality of event occurrences is selected for inclusion in a notification message that is communicated or broadcast to the end user, as well as other users engaged with the information handling system and/or first responders in the affected area.

First claim

1 . A method, in an information handling system comprising a processor and a memory, of predicting an event occurrence based on a question presented to the system, the method comprising:
receiving, by the system, a first question from a first user; generating, by the system, a plurality of context parameters for the first question; generating, by the system, one or more profile parameters for the first user; generating, by the system, one or more historical questions, answers, and events having a specified spatial and/or temporal proximity to the first question; generating, by the system, a ranked plurality of event occurrences based on the plurality of context parameters, the one or more profile parameters for the first user, and the one or more historical questions, answers, and events; and predicting an event occurrence from the ranked plurality of event occurrences based on rank.

2 . The method of claim 1 , wherein generating the plurality of context parameters for the first question comprises performing, by the system, a natural language processing (NLP) analysis of the first question, wherein the NLP analysis extracts key terms, question sentiment, question focus, and lexical answer type information, from the first question.

3 . The method of claim 1 , wherein generating one or more profile parameters for the first user comprises performing, by the system, authorship profile processing analysis of the first user to identify behavioral characteristics of the first user.

4 . The method of claim 1 , wherein generating one or more profile parameters for the first user comprises identifying, by the system, a first user location and time information for when the first question was submitted.

5 . The method of claim 1 , wherein generating one or more historical questions having a specified spatial and/or temporal proximity to the first question comprises performing, by the system, a machine learning approach for identifying one or more historical questions from a question database that are similar to the first question.

6 . The method of claim 1 , wherein generating one or more historical answers having a specified spatial and/or temporal proximity to the first question comprises performing, by the system, a machine learning approach for identifying one or more historical answers from an answer database that are similar to a first answer to the first question.

7 . The method of claim 1 , wherein generating one or more historical events comprises performing, by the system, a probabilistic or vector space model-based search for identifying one or more distress events from an event database that have a specified spatial and/or temporal proximity to the first question.

8 . The method of claim 1 , wherein generating the ranked plurality of event occurrences comprises performing, by the system, a natural language processing (NLP) to execute a machine learning classification model based on model inputs comprising location information for the first user, profile attributes for the first user, submission timing information for the first question, question sentiment, question focus, and lexical answer type information for the first question.

9 . The method of claim 1 , further comprising sending, in near-real time to submission of the first question, a notification message to at least the first user which includes the top ranked event occurrence.

10 . The method of claim 1 , further comprising sending, in near-real time to submission of the first question, a notification message to one or more first responders which includes the predicted event occurrence.

11 . The method of claim 1 , further comprising analyzing, by the system, the first question according to action generation criteria to determine if a threshold number of one or more additional end users are asking related questions within a specified time frame and proximity to the first question.

12 - 20 . (canceled)

Assignees

• IBM

Inventors

• Chandrasekaran Swaminathan
• Dandala Bharath
• Krishnamurthy Lakshminarayanan
• Richardson Alvin C

Classifications

• G06N5/022 Primary

  Knowledge engineering; Knowledge acquisition · CPC title

• G06F17/30401

  Physics · mapped topic

• G06N5/04

  Inference or reasoning models · CPC title

• G06F17/3053 Primary

  Physics · mapped topic

• G06N99/005

  Physics · mapped topic