DETERMINING AT LEAST ONE CATEGORY PATH FOR IDENTIFYING INPUT TEXT

• US 20110112824A1
• Filed: 11/06/2009
• Published: 05/12/2011
• Est. Priority Date: 11/06/2009
• Status: Abandoned Application

• Current Assignee: Hewlett Packard Enterprise Development LP (Hewlett-Packard Enterprise Company)
• Original Assignee: Hewlett Packard Enterprise Development LP (Hewlett-Packard Enterprise Company)
• Inventors: Scholz, Martin, Lukose, Rajan, Zendejas, Ignacio, Sayers, Craig Peter, Rajaram, Shyamsundar
• Application Number:
  US12/614,260
• Publication Number:
  US 20110112824A1
• Time in Patent Office: Days
• US Class Current: 704/9
• CPC Class Codes: G06F 40/237 Lexical tools

Abstract

In a method of determining at least one category path for identifying an input text, one or more categories that are most relevant to the input text are determined, one or more concepts that are most relevant to the input text using information from a labeled text data source and the one or more categories determined to be the most relevant to the input text are determined, and one or more category paths through a hierarchy of predefined category levels are determined for one or more of the determined concepts.

Claims

1. A method of determining at least one category path for identifying an input text, said method comprising:

in a computing device, determining one or more categories that are most relevant to the input text;

determining one or more concepts that are most relevant to the input text using information from a labeled text data source and the one or more categories determined to be the most relevant to the input text; and

determining one or more category paths through a hierarchy of predefined category levels for one or more of the determined concepts.

2. The method according to claim 1, wherein the labeled text data source includes a corpus having a plurality of concepts and categories, said method further comprising:

pre-processing the labeled text data source to find categories for each of the concepts by mapping the categories into a category graph, to find phrases related to each category by using text of articles assigned to the concepts in each category, to find phrases related to each concept by using text anchor tags which point to that concept, and to evaluate counts of occurrences to determine the probability that an occurrence of a particular phrase indicates the text is relevant to a particular category or a particular concept.

3. The method according to claim 2, wherein pre-processing the labeled text data source further comprises creating dictionaries of probabilities that map the concepts to the categories, that map the anchor tags to the categories, and that map the anchor tags to the concepts

4. The method according to claim 3, wherein the labeled text data source comprises a plurality of articles and wherein determining the one or more categories that are most relevant to the input text further comprises comparing the input text with text contained in the plurality of articles by looking up phrases from the input text in the dictionaries and by computing a probability for each of the one or more categories using probabilities for each category based upon whether the phrases from the input text match phrases in the dictionaries.

5. The method according to claim 4, wherein determining at least one of to the one or more categories, the one or more concepts, and the one or more category paths further comprises using information of at least one of a user, a group to which the user belongs, and known about users who are known to be similar to the user.

6. The method according to claim 4, wherein determining the one or more concepts that are most relevant to the input text further comprises comparing the input text with text contained in the plurality of articles to determine which of the concepts is plausibly relevant to the input text by:

searching for phrases from the input text in the dictionaries; and

computing a probability for each concept using the probabilities for each concept based upon whether the phrases from the input text match phrases in the dictionaries and the category probabilities.

7. The method according to claim 6, further comprising:

determining which of the one or more concepts are plausibly relevant to the input text;

determining which of the one or more plausibly relevant concepts are the most relevant to the input text; and

wherein determining the one or more category paths further comprises determining which of the one or more category paths are plausibly relevant to the input text from the determined one or more plausibly relevant concepts.

8. The method according to claim 7, further comprising:

computing metrics for each of the one or more plausibly relevant category paths, wherein the metrics are designed to identify a relevance level for each of the plausibly relevant category paths with respect to the input text, to identify which of the one or more plausibly relevant category paths are the most relevant to the input text.

9. The method according to claim 7, further comprising:

generating at least one category path to identify the input text, wherein the at to least one category path terminates at the one or more plausibly relevant concepts determined to be the most relevant to the input text.

10. An apparatus for determining at least one category path for identifying an input text, said apparatus comprising:

a category determining module configured to determine one or more categories that are most relevant to the input text;

a concept determining module configured to determine one or more concepts that are most relevant to the input text using information from a labeled text data source and the one or more categories determined to be the most relevant to the input text;

a category path determining module configured to determine one or more category paths through a hierarchy of predefined category levels for one or more determined concepts; and

a category path relevance determining module configured to determine which of the one or more category paths is most relevant to the input text.

11. The apparatus according to claim 10, wherein the labeled text data source includes a corpus having a plurality of concepts and categories, said apparatus further comprising:

a pre-processing module configured to pre-process the labeled text data source to find categories for each of the concepts by mapping the categories into a category graph, to find phrases related to each category by using text of articles assigned to the concepts in each category, to find phrases related to each concept by using text anchor tags which point to that concept, and to evaluate counts of occurrences to determine the probability that an occurrence of a particular phrase indicates the text is relevant to a particular category or a particular concept.

12. The apparatus according to claim 11, wherein the pre-processing module is further configured to create dictionaries of probabilities that map the concepts to the categories, that map the anchor tags to the categories, and that map the anchor tags to the concepts.

13. The apparatus according to claim 12, wherein the labeled text data source comprises a plurality of articles and wherein the category determining module is further configured to compare the input text with text contained in the plurality of articles by looking up phrases from the input text in the dictionaries and by computing a probability for each of the one or more categories using probabilities for each category based upon whether the phrases from the input text match phrases in the dictionaries.

14. The apparatus according to claim 13, wherein at least one of the category determining module, the concept determining module, and the category path determining module is further configured to use information of at least one of a user, a group to which the user belongs, and known about users who are known to be similar to the user.

15. The apparatus according to claim 13, wherein the concept determining module is further configured to search for phrases from the input text in the dictionaries and to compute a probability for each concept using the probabilities for each concept based upon whether the phrases from the input text match phrases in the dictionaries and the category probabilities to determine which of the concepts is plausibly relevant to the input text.

16. The apparatus according to claim 15, wherein the concept determining module is further configured to determine which of the one or more concepts are plausibly relevant to the input text and which of the one or more plausibly relevant concepts are the most relevant to the input text, said apparatus further comprising:

a category path relevance determining module configured to identify which of the one or more category paths are plausibly relevant to the input text from the determined one or more plausibly relevant concepts.

17. The apparatus according to claim 16, wherein the category path relevance determining module is further configured to compute metrics for each of the one or more plausibly relevant category paths, wherein the metrics are designed to identify a relevance level for each of the plausibly relevant category paths with respect to the input text, to identify which of the one or more plausibly relevant category paths are the most relevant to the input text.

18. The apparatus according to claim 16, further comprising:

a category path generating module configured to generate at least one category path to identify the input text, wherein the at least one category path terminates at the one or more plausibly relevant concepts determined to be the most relevant to the input text.

19. A computer readable storage medium on which is embedded one or more computer programs, said one or more computer programs implementing a method of determining at least one category path for identifying an input text, said one or more computer programs comprising a set of instructions for:

determining one or more categories that are most relevant to the input text;

determining one or more concepts that are most relevant to the input text using information from a labeled text data source and the one or more categories determining to be the most relevant to the input text; and

determining one or more category paths through a hierarchy of predefined category levels for one or more of the determined concepts.

20. The computer readable storage medium according to claim 19, said one or more computer programs comprising a set of instructions for:

pre-processing the labeled text data source to find categories for each of the concepts by mapping the categories into a category graph, to find phrases related to each category by using text of articles assigned to the concepts in each category, to find phrases related to each concept by using text anchor tags which point to that concept, and to evaluate counts of occurrences to determine the probability that an occurrence of a particular phrase indicates the text is relevant to a particular category or a particular concept.