Disambiguation of concept classifications using language-specific clues

1. A computer-implemented method of classifying a concept of a data item, comprising:

receiving a data item label for the data item;

analyzing the data item label using a natural language processing tool to generate additional lexical information for the data item label;

building a search query using the additional lexical information for the data item label;

using the search query to search a clue concept index to produce a search result, wherein the clue concept index comprises clue concept records that identify clues for concepts, and wherein the search result identifies clue concept records from the clue concept index that match the search query; and

selecting a concept from the clue concept records identified in the search result as the concept for the data item.

2. The computer-implemented method of claim 1 , wherein:

the additional lexical information for the data item label comprises tokens, wherein each token in the tokens comprises text from the data item label and a lemma of the text from the data item label; and

wherein building the search query comprises building the search query using the text and lemma from each token in the tokens for the data item label.

3. The computer-implemented method of claim 1 , wherein the search result comprises a relevance score for each clue concept record identified in the search result, and further comprising:

modifying the relevance score for each clue concept record identified in the search result based on which field in the clue concept record is matched to the search query in the search to produce a boosted relevance score for each clue concept record identified in the search result; and

using the boosted relevance score to select a concept from the clue concept records identified in the search result as the concept for the data item.

4. The computer-implemented method of claim 3 further comprising:

determining a weighted relevance score for each clue concept record identified in the search result by weighting the boosted relevance score for each clue concept record identified in the search result with a weight determined by a scoring algorithm; and

selecting concepts in the clue concept record with the highest weighted relevance score as concepts for the data item.

5. The computer-implemented method of claim 4 , wherein weights determined by the scoring algorithm are trained by a machine learning model.

6. The computer-implemented method of claim 1 , comprising:

identifying a language of the data item label;

determining whether the language of the data item label is English;

in response to a determination that the language of the data item label is not English, using the search query to search a language-specific clue concept index for the language of the data item label and an English clue concept index;

in response to a determination that the language of the data item label is English, using the search query to search only an English clue concept index.

7. The computer implemented method of claim 1 further comprising, generating the clue concept index by:

receiving concept labels for concepts;

automatically generating lexical clues from the concept labels;

analyzing the lexical clues using a natural language processing tool to generate additional lexical information for the concepts;

generating the clue concept records using the lexical clues and additional lexical information as clues for the concepts; and

incorporating the clue concept records in the clue concept index.

8. A system for classifying a concept of a data item, comprising a data processing system configured to:

receive a data item label for the data item;

analyze the data item label using a natural language processing tool to generate additional lexical information for the data item label;

build a search query using the additional lexical information for the data item label;

use the search query to search a clue concept index to produce a search result, wherein the clue concept index comprises clue concept records that identify clues for concepts, and wherein the search result identifies clue concept records from the clue concept index that match the search query; and

select a concept from the clue concept records identified in the search result as the concept for the data item.

9. The system of claim 8 , wherein:

the additional lexical information for the data item label comprises tokens, wherein each token in the tokens comprises text from the data item label and a lemma of the text from the data item label; and

wherein the data processing system is configured to build the search query using the text and lemma from each token in the tokens for the data item label.

10. The system of claim 8 , wherein the search result comprises a relevance score for each clue concept record identified in the search result, and wherein the data processing system is further configured to:

modify the relevance score for each clue concept record identified in the search result based on which field in the clue concept record is matched to the search query in the search to produce a boosted relevance score for each clue concept record identified in the search result; and

use the boosted relevance score to select the concept from the clue concept records identified in the search result as the concept for the data item.

11. The system of claim 10 , wherein the data processing system is further configured to:

determine a weighted relevance score for each clue concept record identified in the search result by weighting the boosted relevance score for each clue concept record identified in the search result with a weight determined by a scoring algorithm; and

select concepts in the clue concept record with the highest weighted relevance score as concepts for the data item.

12. The system of claim 11 , wherein weights determined by the scoring algorithm are trained by a machine learning model.

13. The system of claim 8 , wherein the data processing system is configured to:

identify a language of the data item label;

determine whether the language of the data item label is English;

in response to a determination that the language of the data item label is not English, use the search query to search a language-specific clue concept index for the language of the data item label and an English clue concept index;

in response to a determination that the language of the data item label is English, use the search query to search only an English clue concept index.

14. The system of claim 8 , wherein the data processing system is further configured to generate the clue concept index by:

receiving concept labels for concepts;

automatically generating lexical clues from the concept labels;

analyzing the lexical clues using a natural language processing tool to generate additional lexical information for the concepts;

generating the clue concept records using the lexical clues and additional lexical information as clues for the concepts; and

incorporating the clue concept records in the clue concept index.

15. A computer program product for classifying a concept of a data item, the computer program product comprising a computer readable storage medium having program instructions embodied therewith, the program instructions executable by a device to cause the device to:

receive a data item label for the data item;

analyze the data item label using a natural language processing tool to generate additional lexical information for the data item label;

build a search query using the additional lexical information for the data item label;

use the search query to search a clue concept index to produce a search result, wherein the clue concept index comprises clue concept records that identify clues for concepts, and wherein the search result identifies clue concept records from the clue concept index that match the search query; and

select a concept from the clue concept records identified in the search result as the concept for the data item.

16. The computer program product of claim 15 , wherein:

the additional lexical information for the data item label comprises tokens, wherein each token in the tokens comprises text from the data item label and a lemma of the text from the data item label; and

wherein the program instructions are executable by the device to cause the device to build the search query using the text and lemma from each token in the tokens for the data item label.

17. The computer program product of claim 15 , wherein the search result comprises a relevance score for each clue concept record identified in the search result, and wherein the program instructions are executable by the device to cause the device to:

modify the relevance score for each clue concept record identified in the search result based on which field in the clue concept record is matched to the search query in the search to produce a boosted relevance score for each clue concept record identified in the search result; and

use the boosted relevance score to select the concept from the clue concept records identified in the search result as the concept for the data item.

18. The computer program product of claim 17 , wherein the program instructions are executable by the device to cause the device to:

determine a weighted relevance score for each clue concept record identified in the search result by weighting the boosted relevance score for each clue concept record identified in the search result with a weight determined by a scoring algorithm; and

select concepts in the clue concept record with the highest weighted relevance score as concepts for the data item.

19. The computer program product of claim 15 , wherein the program instructions are executable by the device to cause the device to:

identify a language of the data item label;

determine whether the language of the data item label is English;

in response to a determination that the language of the data item label is not English, use the search query to search a language-specific clue concept index for the language of the data item label and an English clue concept index;

in response to a determination that the language of the data item label is English, use the search query to search only an English clue concept index.

20. The computer program product of claim 15 , wherein the program instructions are executable by the device to cause the device to generate the clue concept index by:

receiving concept labels for concepts;

automatically generating lexical clues from the concept labels;

analyzing the lexical clues using a natural language processing tool to generate additional lexical information for the concepts;

generating the clue concept records using the lexical clues and additional lexical information as clues for the concepts; and

incorporating the clue concept records in the clue concept index.