Utilizing Word Embeddings for Term Matching in Question Answering Systems

Patent metadata

Field	Value
Publication number	US-2016358094-A1
Application number	US-201514727961-A
Country	US
Kind code	A1
Filing date	Jun 2, 2015
Priority date	Jun 2, 2015
Publication date	Dec 8, 2016
Grant date	—

Abstract

Software that compares vector representations of question terms and passage terms in question answering systems by performing the following steps: (i) receiving a question; (ii) generating a plurality of vectors including a first vector representation of a term in the question and a second vector representation of a term in a set of natural language text; (iii) generating a similarity score representing an amount of similarity between the first vector representation and the second vector representation; and (iv) determining whether the set of natural language text is relevant to the question based, at least in part, on the generated similarity score.

First claim

1 - 9 . (canceled)

10 . A computer program product comprising a computer readable storage medium having stored thereon:
first program instructions programmed to receive a question; second program instructions programmed to generate a plurality of vectors including a first vector representation of a term in the question and a second vector representation of a term in a set of natural language text; third program instructions programmed to generate a similarity score representing an amount of similarity between the first vector representation and the second vector representation; and fourth program instructions programmed to determine whether the set of natural language text is relevant to the question based, at least in part, on the generated similarity score.

11 . The computer program product of claim 10 , wherein the generating of the similarity score utilizes unsupervised learning method(s).

12 . The computer program product of claim 11 , wherein the unsupervised learning method(s) include comparing the first vector representation and the second vector representation using a similarity function.

13 . The computer program product of claim 12 , wherein the similarity function is at least one of a cosine similarity function and a Euclidean distance function.

14 . The computer program product of claim 10 , wherein the generating of the similarity score utilizes supervised learning method(s).

15 . The computer program product of claim 14 , wherein the supervised learning method(s) include utilizing an artificial neural network.

16 . A computer system comprising:
a processor(s) set; and a computer readable storage medium; wherein: the processor set is structured, located, connected and/or programmed to run program instructions stored on the computer readable storage medium; and the program instructions include:
first program instructions programmed to receive a question;
second program instructions programmed to generate a plurality of vectors including a first vector representation of a term in the question and a second vector representation of a term in a set of natural language text;
third program instructions programmed to generate a similarity score representing an amount of similarity between the first vector representation and the second vector representation; and
fourth program instructions programmed to determine whether the set of natural language text is relevant to the question based, at least in part, on the generated similarity score.

17 . The computer system of claim 16 , wherein the generating of the similarity score utilizes unsupervised learning method(s).

18 . The computer system of claim 17 , wherein:
the unsupervised learning method(s) include comparing the first vector representation and the second vector representation using a similarity function; and the similarity function is at least one of a cosine similarity function and a Euclidean distance function.

19 . The computer system of claim 16 , wherein the generating of the similarity score utilizes supervised learning method(s).

20 . The computer system of claim 19 , wherein the supervised learning method(s) include utilizing an artificial neural network.

Assignees

• IBM

Inventors

• Fan James J
• Wang Chang
• Xiang Bing
• Zhou Bowen

Classifications

• G06N7/01

  Probabilistic graphical models, e.g. probabilistic networks · CPC title

• G06F40/30

  Semantic analysis · CPC title

• G06F16/3329

  Natural language query formulation · CPC title

• G06N3/084

  Backpropagation, e.g. using gradient descent · CPC title

• G06F16/3344 Primary

  using natural language analysis · CPC title