System and method for personalized search

• US 7,599,916 B2
• Filed: 04/20/2005
• Issued: 10/06/2009
• Est. Priority Date: 04/20/2005
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Weare, Christopher B.
• Primary Examiner(s): Wong; Don
• Assistant Examiner(s): DANG, THANH HA T
• Application Number:
  US11/109,950
• Publication Number:
  US 20060242135A1
• Time in Patent Office: 1,630 Days
• Field of Search: 707 1- 5
• US Class Current: 1/1
• CPC Class Codes: G06F 16/9535 Search customisation based on user profiles and personalisation
• 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 707/99934 Query formulation, input preparation, or translation
• CPC Class Codes: Y10S 707/99935 Query augmenting and refining, e.g. inexact access

Abstract

A system and method for incorporating user preferences into a document search is provided. User events are associated with one or more content tags based on a comparison of the user event with locations in a content space. The user events are tracked to provide a user event history. The user event history is analyzed to select content tags that correspond to user preferences. A search query is modified to include one or more selected content tags. The search query is then matched to documents associated with content tags contained in the search query.

Claims

1. A computer implemented method of performing a search, comprising:

associating a plurality of user events for a user with content tags according to a pre-searching procedure, wherein the pre-searching procedure comprises;

constructing feature vectors for one or more documents that are searched within a document collection, the feature vectors are determined based on a frequency of occurrence of a word or word phrase within the one or more documents of the document collection;

providing a content space grid comprising content tiles, wherein each of the content tiles is associated with a grid feature vector of a plurality of grid feature vectors and is arranged within the content space grid according to subject matter of words or word phrases represented by each grid feature vector of the plurality of grid feature vectors, wherein the arrangement ensures that the content tiles shared a common boundary in the content space grid have related content associated therewith, and wherein the content tags are assigned to each of the content tiles, respectively, and identify locations of the content tiles within the content space grid;

associating the content tiles with the one or more documents in the document collection by comparing the feature vectors of the one or more documents and the plurality of grid feature vectors of the content tiles and, based on the comparison, identifying the feature vectors corresponding to one or more grid feature vectors of the plurality of grid feature vectors;

tracking the plurality of user events affecting the one or more documents; and

associating the content tags with the one or more documents in accordance with the associated content tiles and updating the content tags associated with the one or more documents upon determining that the one or more documents are accessed by the tracked plurality of user events;

receiving a search query from the user;

selecting one or more of the associated content tags according to characteristics of the plurality of user events;

modifying the received search query with the one or more selected content tags;

comparing the modified received search query against the one or more documents of the document collection to find a match; and

at least temporarily storing the one or more matching documents in association with the modified received search query on computer storage medium.

2. The computer implemented method of claim 1, wherein selecting one or more of the associated content tags comprises:

maintaining a queue of the content tags associated with the user events invoked by the user;

determining a threshold position in the queue based on a time-frame when the user events were harvested; and

selecting the content tags within the queue, having a position less than the threshold position.

3. The computer implemented method of claim 1, wherein selecting one or more of the associated content tags comprises:

measuring the frequency of occurrence by which user events of the plurality of user events are associated with each of the content tags;

determining a threshold frequency value based on the measured frequency of occurrence of the user events; and

selecting the content tags having the frequency of occurrence greater than the threshold frequency value.

4. The computer implemented method of claim 1, wherein selecting one or more of the associated content tags comprises:

maintaining a queue of primary content tags having a first resolution associated with user events for the user, each primary content tag being associated with one of the content tiles;

tracking a frequency value for secondary content tags having a second resolution, each secondary content tag being associated with a plurality of content tags having the first resolution;

determining a threshold preference value;

calculating a preference value for each of the content tiles; and

selecting a primary content tag and an associated secondary content tag for each of the content tiles having a preference value above the threshold value.

5. The computer implemented method of claim 4, wherein the preference value for each of the content tiles is calculated using the formula:

6. The computer implemented method of claim 1, wherein a user event of the plurality of user events corresponds to a user visiting one or more documents of the document collection.

7. The computer implemented method of claim 1, wherein a user event of the plurality of user events corresponds to a search query received from the user.

8. The computer implemented method of claim 1, wherein modifying the received search query with the one or more selected content tags further comprises modifying the received search query with one or more content tags corresponding to the content tiles in the content space grid, wherein the corresponding content tiles share a common boundary with the content tiles associated with the one or more selected content tags.

9. The computer implemented method of claim 1, wherein selecting one or more of the associated content tags comprises:

utilizing an inverted index to identify one or more stored content tags corresponding to each keyword in the received search query; and

selecting one or more associated content tags that match at least one stored content tag.

10. A computer readable medium storing computer executable instructions for performing the method of claim 1.

11. One or more computer-readable media having computer-executable instructions embodied thereon that, when executed, perform a method for performing a document search, the method comprising:

providing a user event history comprising a plurality of stored content tags, wherein providing the user event history comprises;

constructing feature vectors for one or more documents that are searched within a document collection, the feature vectors are determined based on a frequency of occurrence of a word or word phrase within the one or more documents of the document collection;

providing a content space grid comprising content tiles, wherein each of the content tiles is associated with a grid feature vector of a plurality of grid feature vectors and is arranged within the content space grid according to subject matter of words or word phrases represented by the grid feature vector of the plurality of grid feature vectors, wherein the arrangement ensures that the content tiles shared a common boundary in the content space grid have related content associated therewith, and wherein the content tags are assigned to each of the content tiles, respectively, and identify locations of the content tiles within the content space grid;

associating the content tags with each of the one or more documents in the document collection according to grid feature vectors, of the plurality of grid feature vectors, associative with the content tiles to which the content tags are assigned;

tracking a plurality of user events affecting the one or more documents; and

storing the content tags associated with the one or more documents indicated by the user event history as the stored content tags;

analyzing the user event history to select one or more of the stored content tags;

modifying a search query to include the one or more selected stored content tags; and

matching the modified search query with the one or more documents utilizing the at least one selected content tag.

12. The one or more computer-readable media of claim 11, wherein the one or more documents each contains a plurality of keywords that, incident to performing the pre-search process, are entered to a document-retrieval inverted index.

13. The one or more computer-readable media of claim 12, further comprising utilizing the document-retrieval inverted index to provide the one or more matching documents in response to the modified search query.

14. The one or more computer-readable media of claim 13, where the one or more matching documents are ordered on a prioritized list, wherein the ordering is based on a number of content tag matches for each of the one or more documents.

15. A computer system for performing context based document searches comprising:

a processing unit coupled to a computer storage medium, the computer storage medium having stored thereon a plurality of computer software components executable by the processing unit, the computer software components comprising a grid builder, a content tag assignment mechanism, a feature association mechanism, a user event tracker, and a keyword matching mechanism;

the grid builder constructs a content space grid comprising content tiles, each of the content tiles is associated with feature vectors and is arranged according to subject matter of words or word phrases represented by the feature vectors, wherein the feature vectors of each of the content tiles are related to a position of the content tiles on the content space grid, and wherein the position ensures that the content tiles shared a common boundary in the content space grid have related words or word phrases associated therewith;

the content tag assignment mechanism assigns a content tag to each of the content tiles according to the feature vectors, wherein the content tag is assigned to each of the content tiles, respectively, and identifies locations of the content tiles within the content space grid;

the feature association mechanism assigns a series of feature values for a document according to a content therein, associating the document with one or more content tiles by comparing the series of feature values for the document against the feature vectors of the content tiles, and assigning the content tag to the document based on the content tiles associated therewith based on the comparison;

the user event tracker maintains a history of user events that indicates documents by updating content tags corresponding to each content tile and selecting the content tags corresponding to user preferences; and

the keyword matching mechanism matches a content tag appended to a search query with a document, wherein matching comprises comparing the content tag appended to the search query against the content tag assigned to each of the one or more content tiles to find at least one corresponding content tile and to identify the document associated therewith.

16. The computer system of claim 15, further comprising a search query modification mechanism to identify a content tag and to modify the search query to include the identified content tag.

17. The computer system of claim 16, wherein identifying a content tag comprises receiving a selected content tag from the user event tracker.

18. The computer system of claim 15, further comprising a document indexing mechanism to store associations between content tags and documents that are assigned according to a pre-search process.

19. The computer system of claim 15, wherein the grid builder further configured to determine feature vectors by identifying words or phrases within training documents, the feature vectors establishing the position of the content tiles on the grid.

20. The computer system of claim 19, wherein the feature association mechanism further configured to associate the document with one or more content tiles upon determining whether the series of feature values for a document satisfy a similarity threshold value.