Information retrieval based on semantic patterns of queries

• US 20110060733A1
• Filed: 08/30/2010
• Published: 03/10/2011
• Est. Priority Date: 09/04/2009
• Status: Active Grant

• Current Assignee: Alibaba Group Holding Ltd.
• Original Assignee: Alibaba Group Holding Ltd.
• Inventors: Peng, Sheng, Zhang, Qin, Hou, Lei, Sun, Jian
• Granted Patent: US 8,799,275 B2
• Time in Patent Office: Days
• US Class Current: 707/723
• CPC Class Codes: G06F 16/9535 Search customisation based on user profiles and personalisation

Abstract

An information retrieval method includes pre-processing a set of historical query information and processing a user query. Pre-processing a set of historical query information includes determining a plurality of semantic patterns based on a plurality of queries in the set of historical query information; establishing correspondence relationships between the plurality of semantic patterns and a plurality of filtering and ranking operations. Processing a user query comprises receiving the user query; retrieve a plurality of results in response to the user query; determining a semantic pattern that corresponds to the user query; determining a set of filtering and ranking operations that corresponds to the semantic pattern based on the correspondence relationships; and performing the set of filtering and ranking operations on the plurality of results to generate a set of filtered and ranked results.

Claims

1. An information retrieval method, comprising:

pre-processing a set of historical query information, comprising;

determining a plurality of semantic patterns based on a plurality of queries in the set of historical query information;

establishing correspondence relationships between the plurality of semantic patterns and a plurality of filtering and ranking operations; and

processing a user query, comprising;

receiving the user query;

retrieve a plurality of results in response to the user query;

determining a semantic pattern that corresponds to the user query;

determining a set of filtering and ranking operations that corresponds to the semantic pattern based on the correspondence relationships; and

performing the set of filtering and ranking operations on the plurality of results to generate a set of filtered and ranked results.

2. The method of claim 1, wherein determining the plurality of semantic patterns based on the plurality of queries in the historical query information includes:

identifying a plurality of semantic tag sets for the plurality of queries, each semantic tag set comprising a plurality of semantic tags that characterizes a plurality of query terms in a corresponding query;

determining possible semantic patterns that correspond to the plurality of queries in the set of historical records based on the plurality of semantic tag sets; and

selecting from possible semantic patterns the plurality of semantic patterns that correspond to the plurality of queries in the set of historical records, each of the plurality of semantic patterns appearing at a rate that at least meets a predetermined threshold.

3. The method of claim 2, wherein selecting from possible semantic patterns the plurality of semantic patterns that correspond to the plurality of queries in the set of historical records includes:

for each semantic pattern, determining total number of corresponding queries that corresponds to the semantic pattern, and selecting said each semantic pattern if the total number of corresponding queries exceeds a threshold.

4. The method of claim 2, wherein selecting from possible semantic patterns the plurality of semantic patterns that correspond to the plurality of queries in the set of historical records includes:

for each semantic pattern, computing entropy of the semantic pattern; and

discarding the semantic patterns if its entropy does not meet a predetermined threshold value.

5. The method of claim 1, wherein establishing correspondence relationships between the plurality of semantic patterns and a plurality of filtering and ranking operations includes:

determining a plurality of user behavior sets that corresponds to the plurality of semantic patterns;

determining a plurality of user intention attribute sets that correspond to the plurality of user behavior sets;

determining the plurality of filtering and ranking operations based on the plurality of user intention attribute sets.

6. The method of claim 1, wherein the plurality of user intention attributes include a degree of ambiguity attribute.

7. The method of claim 1, wherein the plurality of user intention attributes include an authority requirement attribute.

8. The method of claim 1, wherein the plurality of user intention attributes include a time efficiency requirement attribute.

9. The method of claim 1, wherein the plurality of user intention attributes include a location requirement attribute.

10. The method of claim 1, wherein the plurality of user intention attributes include a volume attribute.

11. An information retrieval system, comprising:

one or more processors configured to;

pre-process a set of historical query information, comprising;

determining a plurality of semantic patterns based on a plurality of queries in the set of historical query information;

establishing correspondence relationships between the plurality of semantic patterns and a plurality of filtering and ranking operations; and

process a user query, comprising;

receiving the user query;

retrieve a plurality of results in response to the user query;

determining a semantic pattern that corresponds to the user query;

determining a set of filtering and ranking operations that corresponds to the semantic pattern based on the correspondence relationships; and

performing the set of filtering and ranking operations on the plurality of results to generate a set of filtered and ranked results; and

one or more memories coupled to the processors, configured to provide the processors with instructions.

12. The system of claim 11, wherein determining the plurality of semantic patterns based on the plurality of queries in the historical query information includes:

identifying a plurality of semantic tag sets for the plurality of queries, each semantic tag set comprising a plurality of semantic tags that characterizes a plurality of query terms in a corresponding query;

determining possible semantic patterns that correspond to the plurality of queries in the set of historical records based on the plurality of semantic tag sets; and

selecting from possible semantic patterns the plurality of semantic patterns that correspond to the plurality of queries in the set of historical records, each of the plurality of semantic patterns appearing at a rate that at least meets a predetermined threshold.

13. The system of claim 12, wherein selecting from possible semantic patterns the plurality of semantic patterns that correspond to the plurality of queries in the set of historical records includes:

for each semantic pattern, determining total number of corresponding queries that corresponds to the semantic pattern, and selecting said each semantic pattern if the total number of corresponding queries exceeds a threshold.

14. The system of claim 12, wherein selecting from possible semantic patterns the plurality of semantic patterns that correspond to the plurality of queries in the set of historical records includes:

for each semantic pattern, computing entropy of the semantic pattern; and

discarding the semantic patterns if its entropy does not meet a predetermined threshold value.

15. The system of claim 11, wherein establishing correspondence relationships between the plurality of semantic patterns and a plurality of filtering and ranking operations includes:

determining a plurality of user behavior sets that corresponds to the plurality of semantic patterns;

determining a plurality of user intention attribute sets that correspond to the plurality of user behavior sets;

determining the plurality of filtering and ranking operations based on the plurality of user intention attribute sets.

16. The system of claim 11, wherein the plurality of user intention attributes include a degree of ambiguity attribute.

17. The system of claim 11, wherein the plurality of user intention attributes include an authority requirement attribute.

18. The system of claim 11, wherein the plurality of user intention attributes include a time efficiency requirement attribute.

19. The system of claim 11, wherein the plurality of user intention attributes include a location requirement attribute.

20. The system of claim 11, wherein the plurality of user intention attributes include a volume attribute.

21. A computer program product for inferring a characteristic of an individual, the computer program product being embodied in a computer readable storage medium and comprising computer instructions for:

pre-processing a set of historical query information, comprising;

determining a plurality of semantic patterns based on a plurality of queries in the set of historical query information;

establishing correspondence relationships between the plurality of semantic patterns and a plurality of filtering and ranking operations; and

processing a user query, comprising;

receiving the user query;

retrieve a plurality of results in response to the user query;

determining a semantic pattern that corresponds to the user query;

determining a set of filtering and ranking operations that corresponds to the semantic pattern based on the correspondence relationships; and

performing the set of filtering and ranking operations on the plurality of results to generate a set of filtered and ranked results.