Demographic based classification for local word wheeling/web search

• US 7,778,837 B2
• Filed: 11/30/2006
• Issued: 08/17/2010
• Est. Priority Date: 05/01/2006
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Thiesson, Bo, Church, Kenneth W.
• Primary Examiner(s): Dorvil; Richemond
• Assistant Examiner(s): Saint Cyr; Leonard
• Application Number:
  US11/565,231
• Publication Number:
  US 20070255552A1
• Time in Patent Office: 1,356 Days
• Field of Search: None
• US Class Current: 704/278
• CPC Class Codes: G06F 16/9535 Search customisation based on user profiles and personalisation
• CPC Class Codes: G06F 16/9537 Spatial or temporal dependent retrieval, e.g. spatiotemporal queries

Abstract

Systems and methods that create a classification of sentences in a language, and further construct associated local versions of language models, based on geographical location and/or other demographic criteria—wherein such local language models can be of different levels of granularity according to chosen demographic criteria. The subject innovation employs a classification encoder component that forms a classification (e.g. a tree structure) of sentences, and a local language models encoder component, which employs the classification of sentences in order to construct the localized language models. A decoder component can subsequently enable local word wheeling and/or local web search by blending k-best answers from local language models of varying demographic granularity that match users demographics. Hence, k-best matches for input data by users in one demographic locality can be different from k-best matches for the same input by other users in another locality.

Claims

1. A computer implemented system comprising the following computer executable components:

multiprocessors;

a classification encoder component that summarizes a query log into a classification based on demographics, wherein the classification is in the form of a tree structure to summarize the query log, wherein the classification encoder component leverages information stored in the query log by selectively mining a plurality of queries from the query log and summarizing the plurality of queries in the tree structure;

a local language models encoder component that expands the classification to form localized language models, to facilitate a search for local interests; and

an artificial intelligence component that facilitates smoothing of the classification, wherein the artificial intelligence component comprises a support vector machine (SVM) configured via a learning phase within a classifier constructor and a feature selection module, wherein the classifier constructor is used to learn and perform a determination according to a predetermined criteria to update a previously inferred schema, modify the criteria on a referring algorithm based at least in part upon the kind of data being processed and select the time of day to implement the modifying of the criteria.

2. The computer implemented system of claim 1, further comprising a decoder component with an expansion component that evaluates input data via classification of local language models to produce a set of expansions.

3. The computer implemented system of claim 1, wherein the tree structure with nodes promotes queries based on predetermined criteria.

4. The computer implemented system of claim 3, wherein the predetermined criteria are significance test criteria.

5. The computer implemented system of claim 1, further comprising a local context component that provides information about a user'"'"'s locality in order to produce expansions relevant to local interests.

6. The computer implemented system of claim 1, wherein the demographics are based on at least one of time and space.

7. The computer implemented system of claim 1, wherein the tree structure is in the form of a Kd-tree structure.

8. The computer implemented system of claim 1 further comprising an interface component that accepts input data from a user.

9. A computer implemented method comprising the following computer executable acts:

summarizing, by a processor, query logs in the form of a tree structure based at least in part on demographics, the summarizing comprising;

leveraging information stored in a query log by selectively mining a plurality of queries from the query log; and

summarizing the plurality of queries in the tree structure;

forming localized language models from the classification; and

smoothing the tree structure, the smoothing comprising;

using an artificial intelligence component, the artificial intelligence component comprising;

configuring a support vector machine (SVM) using a learning phase within a classifier constructor and a feature selection module, wherein the classifier constructor is used to automatically learn and perform a determination according to a predetermined criteria to update a previously inferred schema, modify the criteria on a referring algorithm based upon the kind of data being processed and select the time of day to implement the modifying of the criteria.

10. The computer implemented method of claim 9 wherein the smoothing further comprises moving the sentences among levels of the tree structure.

11. The computer implemented method of claim 10 further comprising expanding an input based on the localized language.

12. The computer implemented method of claim 9, the forming act comprising forming local language model(s) for the nodes.

13. The computer implemented method of claim 12 further comprising promoting nodes based on predetermined significance criteria.

14. The computer implemented method of claim 12 further comprising obtaining k best matches in every node on a path from a leaf node to a root node of the tree, and combining the k best matches.

15. The computer implemented method of claim 9, wherein the smoothing is based on predetermined criteria.

16. A computer implemented system comprising the following computer executable components:

multiprocessors;

means for forming a classification structure of local language models based on demographics, the forming comprising a classification encoder component, wherein the classification is in the form of a tree structure to summarize query logs, wherein the classification encoder component leverages information stored in a query log via selectively mining a plurality of queries from the query log and summarizing the plurality of queries in the tree structure;

means for creating localized language models from the classification; and

means for employing the classification structure of local language models to facilitate input expansions of local interest, the employing comprising an artificial intelligence component, wherein the artificial intelligence component comprises a support vector machine (SVM) configured via a learning phase within a classifier constructor and a feature selection module, wherein the classifier constructor is used to automatically learn and perform a determination according to a predetermined criteria to update a previously inferred schema, modify the criteria on a referring algorithm based upon the kind of data being processed and select the time of day to implement the modifying of the criteria.