# US 2019/0171792 A1 · Interaction network inference from vector representation of words

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2019/0171792 A1
- Document ID: US20190171792A1
- Status: Published application
- Kind: A1
- Application number: 15/829,123
- Filed: 2017-12-01
- Published: 2019-06-06
- Earliest claimed: 2017-12-01
- Assignees: International Business Machines Corporation
- Applicants: International Business Machines Corporation
- Inventors: Matteo Manica; Roland Mathis; Maria Rodriguez Martinez; Konstantinos Bekas
- Web page: https://patentgrep.com/patent/US20190171792A1

## CPC

- G06F 16/355: Creation or modification of classes or clusters
- G06F 18/22: Matching criteria, e.g. proximity measures
- G06F 18/23213: with fixed number of clusters, e.g. K-means clustering
- G06F 18/24147: {Distances to closest patterns, e.g. nearest neighbour classification}
- G06F 40/216: using statistical methods
- G06F 40/284: Lexical analysis, e.g. tokenisation or collocates
- G06F 40/30: Semantic analysis
- G16B 15/00: ICT specially adapted for analysing two-dimensional [2D] or three-dimensional [3D] molecular structures, e.g. structural or functional relations or structure alignment
- G16B 40/00: ICT specially adapted for biostatistics; ICT specially adapted for bioinformatics-related machine learning or data mining, e.g. knowledge discovery or pattern finding
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G16B: BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY

## Abstract

The present disclosure relates to a system and method for extracting information from text data. The method comprises: obtaining a plurality of text elements. A word embedding algorithm may be applied to the obtained text elements by mapping each text element of at least part of the text elements into a vector of a predefined dimension. The mapped text elements may be clustered into groups using the distances between the respective vectors. For each text element of a set of text elements of the mapped text elements a respective distribution of neighbors across the groups may be built. Similarity scores may be computed using the distributions thereby for determining relations between the set of text elements.

## Claims

### Claim 1

A computer implemented method to extract information from text data, the method comprising:
obtaining a plurality of text elements;
applying a word embedding algorithm to the obtained text elements comprising mapping each text element of at least part of the text elements into a vector of a predefined dimension;
clustering the mapped text elements into groups using the distances between the respective vectors;
building for a set of text elements of the mapped text elements a respective distribution of neighbors across the groups;
computing similarity scores using the distributions thereby determining relations between the set of text elements.

### Claim 2 (depends on 1)

The method of claim 1, the set of text elements comprising names of molecular entities, the method further comprising building a molecular interaction network using the similarity scores.

### Claim 3 (depends on 1)

The method of claim 1, further comprising providing a vocabulary of text elements, wherein the set of text elements are present in both the mapped text elements and the vocabulary.

### Claim 4 (depends on 1)

The method of claim 1, the building of the distribution for a given text element comprising: identifying a predefined number of nearest neighbors of the given text element;
determining the group to which belongs each of the identified nearest neighbors, wherein the distribution for the given text element indicates for each determined group the fraction of nearest neighbors of the identified nearest neighbors that belongs to the determined group over the predefined number of nearest neighbors.

### Claim 5 (depends on 4)

The method of claim 4, wherein identifying the nearest neighbors comprises building a k-d tree of the vectors representing the text elements and using the k-d tree for identifying the nearest neighbors.

### Claim 6 (depends on 4)

The method of claim 4, wherein the predefined number of nearest neighbors is determined based on the number of groups.

### Claim 7 (depends on 1)

The method of claim 1, the obtaining of the text data comprising:
querying a predefined database for one or more documents;
parsing the queried documents for extracting text;
identifying in the extracted text at least one of bi-grams, trigrams, words and a combination thereof for obtaining the text elements.

### Claim 8 (depends on 1)

The method of claim 1, wherein the similarity scores are Jensen-Shannon divergence, JSD, scores which are computed for pairs of the built distributions.

### Claim 9 (depends on 1)

The method of claim 1, wherein the clustering is performed using a K-means algorithm.

### Claim 10 (depends on 1)

The method of claim 1, wherein the text elements form un-structured data.

### Claim 11 (depends on 1)

The method of claim 1, being automatically performed.

### Claim 12

A computer program product comprising a computer-readable storage medium having computer-readable program code embodied therewith, the computer-readable program code configuring a computer system to:
obtain a plurality of text elements;
apply a word embedding algorithm to the obtained text elements comprising mapping each text element of at least part of the text elements into a vector of a predefined dimension;
cluster the mapped text elements into groups using the distances between the respective vectors;
build for a set of text elements of the mapped text elements a respective distribution of neighbors across the groups; and
compute similarity scores using the distributions thereby determining relations between the set of text elements.

### Claim 13

A computer system to extract information from text data, the computer system comprising:
a processor; and
a non-transitory computer-readable memory coupled to the processor, wherein the memory comprises instructions which, when executed by the processor, cause the processor to:
obtain a plurality of text elements;
apply a word embedding algorithm to the obtained text elements comprising mapping each text element of at least part of the text elements into a vector of a predefined dimension;
cluster the mapped text elements into groups using the distances between the respective vectors;
build for a set of text elements of the mapped text elements a respective distribution of neighbors across the groups;
compute similarity scores using the distributions thereby determining relations between the set of text elements.

### Claim 14 (depends on 13)

The system of claim 13, the set of text elements comprising names of molecular entities, the method further comprising building a molecular interaction network using the similarity scores.

### Claim 15 (depends on 13)

The system of claim 13, wherein the set of text elements are present in both the mapped text elements and a vocabulary of text elements.

### Claim 16 (depends on 13)

The system of claim 13, wherein the processor is configured to perform the building of the distribution for a given text element by: identifying a predefined number of nearest neighbors of the given text element; determining the group to which belongs each of the identified nearest neighbors, wherein the distribution for the given text element indicates for each determined group the fraction of nearest neighbors of the identified nearest neighbors that belongs to the determined group over the predefined number of nearest neighbors.

### Claim 17 (depends on 16)

The system of claim 16, wherein the processor is configured for identifying the nearest neighbors by building a k-d tree of the vectors representing the text elements and using the k-d tree for identifying the nearest neighbors.

### Claim 18 (depends on 16)

The system of claim 16, wherein the processor is configured to determine the predefined number of nearest neighbors based on the number of groups.

### Claim 19 (depends on 13)

The system of claim 13, wherein the processor is configured for obtaining of the text data by:
querying a predefined database for one or more documents;
parsing the queried documents for extracting text;
identifying in the extracted text at least one of bi-grams, trigrams, words and a combination thereof for obtaining the text elements.

### Claim 20 (depends on 13)

The system of claim 13, wherein the similarity scores are Jensen-Shannon divergence, JSD, scores which are computed for pairs of the built distributions.
