Method for disambiguated features in unstructured text

Patent metadata

Field	Value
Publication number	US-2016110446-A1
Application number	US-201514979703-A
Country	US
Kind code	A1
Filing date	Dec 28, 2015
Priority date	Dec 2, 2013
Publication date	Apr 21, 2016
Grant date	—

Abstract

A method for disambiguating features in unstructured text is provided. The disclosed method may not require pre-existing links to be present. The method for disambiguating features in unstructured text may use co-occurring features derived from both the source document and a large document corpus. The disclosed method may include multiple modules, including a linking module for linking the derived features from the source document to the co-occurring features of an existing knowledge base. The disclosed method for disambiguating features may allow identifying unique entities from a knowledge base that includes entities with a unique set of co-occurring features, which in turn may allow for increased precision in knowledge discovery and search results, employing advanced analytical methods over a massive corpus, employing a combination of entities, co-occurring entities, topic IDs, and other derived features.

First claim

What is claimed is:

1 . A method comprising:
in response to receiving, by a server, a search query from a client:
searching, by the server, a set of records comprising a co-occurring feature, wherein the server comprises a main memory hosting a database, wherein the database stores a first cluster, wherein the first cluster comprises a disambiguated primary feature with a unique identifier and a set of secondary features, wherein the first cluster comprises a first score;
identifying, by the server, a record in the set of records, wherein the record matches an extracted feature such that the extracted feature is a primary feature;
associating, by the server, the extracted feature with a topic identifier;
disambiguating, by the server, the primary feature based on a relatedness of the topic identifier;
identifying, by the server, the set of secondary features based on the relatedness;
disambiguating, by the server, the primary feature from the set of secondary features based on the relatedness;
accessing, by the server, the database;
linking, by the server, in real-time, during the accessing, the primary feature to the set of secondary features;
forming, by the server, a second cluster based on the linking, wherein the second cluster comprises a second score;
comparing, by the server, the first score against the second score; determining, by the server, whether the first score matches the second score; identifying, by the server, the unique identifier related to the primary feature in the first cluster based on the first score matching the second score; amending, by the server, based on the identifying the unique identifier, the first cluster such that the first cluster includes the second cluster; and sending, by the server, the unique identifier to the client.

2 . The method of claim 1 , further comprising:
comparing, by the server, each member of the set of records which matches the extracted feature against a data item; assigning, by the server, a third score to the extracted feature based on the comparing of the each of the member.

3 . The method of claim 2 , further comprising:
associating, by the server, the extracted feature with a feature attribute.

4 . The method of claim 3 , wherein the feature attribute is weighted.

5 . The method of claim 2 , further comprising:
determining, by the server, a relatedness of the extracted feature based on the feature attribute.

6 . The method of claim 1 , wherein the primary feature is associated with a feature attribute.

7 . The method of claim 1 , wherein the extracted feature is associated with a lower-ordinal feature in accordance with a cluster hierarchy.

8 . The method of claim 1 , wherein the searching is in a fuzzy manner.

9 . The method of claim 1 , further comprising:
comparing, by the server, a first feature against a second feature, wherein the first feature comprises the extracted feature, wherein the first feature is provided via a first data source, wherein the second feature is provided via a second data source;
determining, by the server, if the first feature co-occurs in the second data source based on the comparing of the first feature against the second feature;
linking, by the server, at least one of the first data source or the second data source.

10 . The method of claim 1 , further comprising:
determining, by the server, a co-occurrence of the extracted feature in a plurality of data sources; improving, by the server, a rate of accuracy of the disambiguating based on the determining of the co-occurrence of the extracted feature.

11 . A method comprising:
in response to receiving, by a server, a search query from a client:
searching, by the server, based on the receiving, a set of records comprising a co-occurring feature, wherein the server comprises a main memory hosting a database, wherein the database stores a first cluster, wherein the first cluster comprises a disambiguated primary feature with a first unique identifier and a set of secondary features, wherein the first cluster comprises a first score;
identifying, by the server, a record in the set of records, wherein the record matches an extracted feature such that the extracted feature is a first primary feature;
associating, by the server, the extracted feature with a topic identifier;
disambiguating, by the server, the first primary feature based on a relatedness of the topic identifier;
identifying, by the server, the set of secondary features based on the relatedness;
disambiguating, by the server, the first primary feature from the set of secondary features based on the relatedness;
accessing, by the server, the database;
linking, by the server, in real-time, during the accessing, the first primary feature to the set of secondary features;
forming, by the server, a second cluster based on the linking, wherein the second cluster comprises a second score;
comparing, by the server, the first score against the second score; determining, by the server, whether the first score matches the second score; generating, by the server, a third cluster based on the first score not matching the second score, wherein the third cluster comprises a second primary feature; assigning, by the server, a second unique identifier to the second primary feature; sending, by the server, the second unique identifier to the client.

12 . The method of claim 1 , further comprising:
comparing, by the server, each member of the set of records which matches the extracted feature against a data item; assigning, by the server, a third score to the extracted feature based on the comparing of the each of the member.

13 . The method of claim 2 , further comprising:
associating, by the server, the extracted feature with a feature attribute.

14 . The method of claim 3 , wherein the feature attribute is weighted.

15 . The method of claim 2 , further comprising:
determining, by the server, a relatedness of the extracted feature based on the feature attribute.

16 . The method of claim 1 , wherein at least one of the first primary feature or the second primary feature is associated with a feature attribute.

17 . The method of claim 1 , wherein the extracted feature is associated with a lower-ordinal feature in accordance with a cluster hierarchy.

18 . The method of claim 1 , wherein the searching is in a fuzzy manner.

19 . The method of claim 1 , further comprising:
comparing, by the server, a first feature against a second feature, wherein the first feature comprises the extracted feature, wherein the first feature is provided via a first data source, wherein the second feature is provided via a second data source;
determining, by the server, if the first feature co-occurs in the second data source based on the comparing of the first feature against the second feature;
linking, by the server, at least one of the first data source or the second data source.

20 . The method of claim 1 , further comprising:
determining, by the server, a co-occurrence of the extracted feature in a plurality of data sources; improving, by the server, a rate of accuracy of the disambiguating based on the determining of the co-occurrence of the extracted feature.

Assignees

• Qbase Llc

Inventors

• Lightner Scott
• Weckesser Franz
• Boddhu Sanjay
• Dave Rakesh
• Flagg Robert

Classifications

• G16B40/00

  ICT specially adapted for biostatistics; ICT specially adapted for bioinformatics-related machine learning or data mining, e.g. knowledge discovery or pattern finding · CPC title

• G06F16/3344

  using natural language analysis · CPC title

• G06F40/279

  Recognition of textual entities · CPC title

• G16C20/70

  Machine learning, data mining or chemometrics · CPC title

• G06F16/35 Primary

  Clustering; Classification · CPC title