Method for measuring thresholded relevance of a document to a specified topic

• US 5,778,363 A
• Filed: 12/30/1996
• Issued: 07/07/1998
• Est. Priority Date: 12/30/1996
• Status: Expired due to Term

• Current Assignee: Intel Corporation
• Original Assignee: Intel Corporation
• Inventors: Light, John
• Primary Examiner(s): Kulik, Paul V.
• Application Number:
  US08/774,467
• Time in Patent Office: 554 Days
• Field of Search: 707/1, 707/3, 707/6, 707/5, 707/513
• US Class Current: 1/1
• CPC Class Codes: G06F 16/335 Filtering based on additional data, e.g. user or group profiles filtering in web context G06F16/9535, G06F16/9536
• CPC Class Codes: Y10S 707/99931 Database or file accessing
• CPC Class Codes: Y10S 707/99933 Query processing, i.e. searching
• CPC Class Codes: Y10S 707/99935 Query augmenting and refining, e.g. inexact access

Abstract

A method is provided for specifying the representation of a document and determining the relevance of the document according to an externally defined topic profile. The topic profile includes one or more compound terms having a positive correlation with the topic of interest. Each compound term has a specified form such as capitalization, punctuation, number, or adjacency relation, that is either ignored by conventional indexing processes or requires substantial data overhead to track. The compound terms of the topic profile are tagged to indicate how corresponding terms are treated when identified in a document being analyzed. Application of the topic profile to a document generates a document representation in which compound terms present in the document are retained in their specified form. A similarity function between the document representation and the topic profile is calculated, and the result is compared to a relevance threshold associated with the topic profile. A document is deemed relevant to the topic when the similarity function meets or exceeds the threshold.

Claims

1. A method for determining a thresholded measure of relevance of a document to a topic defined by a topic profile that includes a compound term template, the method comprising the steps of:

converting the document into a stream of tokens;

scanning the stream for one or more tokens matching the compound term template;

augmenting the document tokens with a compound term token to provide a representation of the document, when a match is detected; and

calculating a similarity transform between the document tokens and the topic profile to provide the measure of relevance.

2. The method of claim 1, including the additional step of eliminating selected tokeins from the document representation, following the augmenting step.

3. The method of claim 2, wherein the eliminating step comprises the substeps of:

stopping selected tokens in the document representation; and

stemming selected tokens in the document representation.

4. The method of claim 1, wherein the calculating step comprises the substeps of:

determining whether the document representation includes a by-pass profile term, when a by-pass profile term is specified in the topic profile;

determining a threshold relevance according to a tag associated with the by-pass profile term, when the

5. A method for identifying a topic to be searched in a document that is to be represented as a plurality of tokens, the method comprising the steps of:

identifying one or more profile terms having a positive correlation with the topic;

selecting from the one or more profile terms a compound term to be identified in a specified form in the document; and

tagging the compound term to indicate a retention status for a token assigned to the compound term in the document representation when an instance of the compound term is identified in the document.

6. The method of claim 5, wherein the selecting step comprises selecting as the specified form a specified capitalization of the profile term.

7. The method of claim 5, wherein the selecting step comprises selecting as the specified form a specified punctuation for the profile term.

8. The method of claim 5, wherein the selecting step comprises selecting as the specified form a specified plural/singular form for the profile term.

9. The method of claim 5, wherein the selecting step comprises selecting as the specified form a specified location configuration for two or more profile terms.

10. The method of claim 9, wherein the tagging step comprises tagging the compound term to indicate that tokens for the two or more profile terms and the token assigned to the compound term be retained in the representation of the document.

11. The method of claim 9, wherein the tagging step comprises tagging the compound term to indicate that tokens for the two or more profile terms be excluded from the representation of the document.

12. The method of claim 5, comprising the additional steps of:

filtering the plurality of tokens representing the document in accordance with the retention status indicated for the tagged compound terms;

stopping and stemming the filtered plurality of tokens to generate a document representation for the document set; and

calculating a similarity function between the document representation and the profile terms.

13. The method of claim 5, comprising the additional step of assigning a set of weights to the one or more profile terms to indicate the relative significance of the profile terms to the identified topic.

14. The method of claim 13, wherein the step of assigning weights comprises the substeps of:

assigning a first weight to each of the one or more profile terms for calculating a threshold relevance for the document; and

assigning a second weight to each of the one or more profile terms for calculating a comparative relevance for the document.

15. The method of claim 14, comprising the additional steps of:

filtering the plurality of tokens representing the document in accordance with the retention status indicated for the tagged compound terms;

stopping and stemming the filtered plurality of tokens to generate a document representation for the document set; and

calculating a similarity function between the document representation and the profile terms using the first set of weights.

16. The method of claim 15, comprising the additional steps of:

deeming the document relevant to the topic when the calculated similarity function exceeds a threshold value associated with the topic profile; and

calculating a similarity function between the document representation and the topic profile using the second set of weights when the document is deemed relevant to the topic.

17. The method of claim 5, comprising the additional step of tagging selected profile terms, including the selected compound term, to indicate that the selected term must be present in order to deem a document in the document set relevant to the topic.

18. The method of claim 5, comprising the additional step of tagging selected profile terms to indicate that presence of the selected profile term in the document is sufficient to deem the document relevant to the topic.

19. The method of claim 5, comprising the additional step of tagging selected profile term to indicate that the presence of the selected profile term in the document is required to deem the document relevant to the topic.

20. A method for generating a representation of a document as a plurality of tokens tailored for searches on a selected topic, the method comprising the steps of:

identifying one or more profile terms having positive correlations with the selected topic;

selecting a profile term to be identified in a specified form in the one or more documents; and

tagging the selected profile term to indicate a retention status for an associated token in the representation of the one or more documents.

21. A method for determining a measure of relevance of a document, represented as a plurality of tokens, to a specified subject area, the method comprising the steps of:

identifying a plurality of topics within the specified subject area;

for each identified topic, selecting a topic profile that includes one or more compound terms that are characteristic of the topic, have a specified form to be identified in the plurality of tokens representing the document, and an associated retention status tag;

modifying the plurality of tokens representing the document according to associated retention status tags when instances of the one or more compound terms are identified in the plurality of tokens;

stopping and stemming the modified plurality of tokens to generate a document representation; and

calculating a similarity measure between each of the plurality of topic profiles and the document representation.

22. The method of claim 21, comprising the additional steps of:

identifying a relevance threshold associated with each of the topic profiles;

comparing the calculated similarity measure for a topic profile to the associated relevance threshold; and

adding an indication of not relevant to the measure of relevance for the topic profile when the calculated similarity measure is less than the associated relevance threshold.

23. The method of claim 22, comprising the additional step of adding the calculated similarity measure for a topic profile to the measure of relevance when the calculated similarity measure is at least as large as the associated relevance threshold.

24. The method of claim 21, wherein the step of calculating a similarity measure comprises the substeps of:

identifying each topic profile term present in the document representation; and

adding a weight associated with each identified profile term to a sum that represents the similarity measure.

25. A topic profile for evaluating the relevance to a topic of a document that has been tokenized into a stream of terms, the topic profile comprising:

one or more profile terms that are characteristic of the topic;

one or more tags associated with selected profile terms to indicate retention/elimination of document terms represented by the selected profile terms in the stream; and

a relevance threshold, wherein the topic profile is applied to tokenized terms of the document to preserve selected profile terms in a document representation and he one or more compound terms

26. A method for generating a topic profile to determine a measure of relevance of a document to a topic characterized by the topic profile, the method comprising the steps of:

identifying a term used to discuss the topic;

eliminating the identified term when it is a common term;

adding the term the topic profile when the term has a strong, positive correlation with the topic and does not alias other topics; and

when a term aliases other topics, defining a specified form for the term that reduces aliasing and maintains a positive correlation with the topic, and adding the specified form to the topic profile as a compound term.

27. The method of claim 26, wherein the step of defining a specified form comprises specifying a capitalization for the term.

28. The method of claim 26, wherein the step of defining a specified form comprises specifying punctuation for the term.

29. The method of claim 26, wherein the step of defining a specified form comprises specifying a singular/plural form for the term.

30. The method of claim 27, wherein the step of defining a specified form comprises specifying a location configuration with one or more other terms.