RECOMMENDATION OF SEARCH KEYWORDS BASED ON INDICATION OF USER INTENTION

• US 20120173562A1
• Filed: 12/22/2011
• Published: 07/05/2012
• Est. Priority Date: 12/31/2010
• Status: Active Grant

• Current Assignee: Alibaba Group Holding Ltd.
• Original Assignee: Alibaba Group Holding Ltd.
• Inventors: Zhu, Li, Zhu, Xiaocong
• Granted Patent: US 8,799,306 B2
• Time in Patent Office: Days
• US Class Current: 707/767
• CPC Class Codes: G06F 16/3322 using system suggestions G06F16/3325 takes precedence
• CPC Class Codes: G06F 16/90324 using system suggestions

Abstract

Search keyword recommendation includes: receiving a set of search keywords input by a user; determining whether the set of search keywords indicate a clearly defined intention by the user, including comparing the set of search keywords with a plurality of predetermined words corresponding to intentions that are not clearly defined; in the event that the set of search keywords are determined not to indicate a clearly defined intention, executing a first recommendation method to obtain recommended search keywords; wherein: the first recommendation method is selected among a plurality of recommendation methods; and the first recommendation method includes a knowledge-based recommendation method, a session correlation-based method, or a combination.

Claims

1. A method comprising:

receiving a set of search keywords input by a user;

determining whether the set of search keywords indicate a clearly defined intention by the user, including comparing the set of search keywords with a plurality of predetermined words corresponding to intentions that are not clearly defined; and

in the event that the set of search keywords are determined not to indicate a clearly defined intention, executing a first recommendation method to obtain recommended search keywords;

wherein;

the first recommendation method is selected among a plurality of recommendation methods; and

the first recommendation method includes a knowledge-based recommendation method, a session correlation-based method, or a combination.

2. The method of claim 1, in the event that the set of search keywords are determined to is indicate a clearly defined intention, further comprising executing a second recommendation method to obtain additional recommended search keywords.

3. The method of claim 2, wherein the second recommendation method includes a search log-based recommendation method.

4. The method of claim 1, wherein:

the first recommendation method includes a combination of at least a knowledge-based recommendation method and a session correlation-based recommendation method; and

the knowledge-based recommendation method contributes to a first proportion of recommended keywords and the session correlation-based recommendation method contributes to a second proportion of recommended keywords.

5. The method of claim 4, further comprising:

recording a first exposure to feedback rate of the knowledge-based recommendation method and a second exposure to feedback rate of the session correlation-based recommendation method; and

adjusting the first proportion of recommended keywords contributed by the knowledge-based recommendation method for future received search keyword sets, and the second proportion of recommended keywords contributed by the session correlation-based recommendation method for future received search keyword sets.

6. The method of claim 1, wherein the knowledge-based recommendation method comprises:

obtaining a plurality of lexical item sets based on search logs;

selecting a lexical item set from the plurality of lexical item sets, the selected lexical item set includes one or more lexical items that best match the set of search keywords;

selecting a selected rule from a stored rule set, wherein each rule in the stored rule set includes a priority-sequenced chain of lexical item types, and in the selected rule, the lexical item type of the selected lexical item set is the first item in the corresponding priority-sequenced chain;

identifying a second lexical item type in the priority-sequenced chain in the selected rule; and

determining a lexical item that is in the set of the second lexical item type and that logically corresponds to the received search keyword set as one of the recommended search keywords.

7. The method of claim 6, wherein the plurality of lexical item sets include a compound-type lexical item set, a mono-type lexical item set, a brand-type lexical item set, and a product model-type lexical item set.

8. The method of claim 1, wherein the session correlation-based recommendation method comprises:

storing search keyword chains corresponding to search sessions by users;

identifying a plurality of search keyword sets listed after the received search keyword set in the stored search keyword chain comprising the received search keyword set; and

determining the recommended search keywords based on positions of the plurality of search keyword sets.

9. The method of claim 8, wherein determining the recommended search keywords includes:

based on pre-stored sets having lexical items of various categories of industries, selecting from the search keyword sets listed after the input search keywords in the search keyword chains a set of search keywords whose category of industry is the same as that of the received search keyword as recommended search keywords.

10. The method of claim 8, wherein determining the recommended search keywords includes:

calculating probabilities of different later-listed search keyword sets occurring among all later-listed search keyword sets; and

selecting, based on the probabilities, search keyword sets that are most likely to occur as recommended search keywords.

11. The method of claim 1, further comprising:

determining, based on pre-stored search keyword sets and their corresponding categories of industry, related categories of industry corresponding to the received search keyword set;

determining click factor attribute values of the received search keyword set for each is related category of industry;

determining probabilities of the received search keyword set belonging to the related categories of industry based at least in part on the determined click factor attribute values;

determining numbers of recommended keywords corresponding to the related categories of industry based on the probabilities; and

wherein executing the first recommendation method to obtain recommended search keywords includes determining recommended search keywords for the related category of industry.

12. The method of claim 1, further comprising the set of search keywords with a plurality of predetermined words corresponding to intentions that are clearly defined.

23. The system of claim 12, further comprising:

determining, based on pre-stored search keyword sets and their corresponding categories of industry, related categories of industry corresponding to the received search keyword set;

determining click factor attribute values of the received search keyword set for each related category of industry;

determining probabilities of the received search keyword set belonging to the related categories of industry based at least in part on the determined click factor attribute values;

determining numbers of recommended keywords corresponding to the related categories of industry based on the probabilities; and

wherein executing the first recommendation method to obtain recommended search keywords includes determining recommended search keywords for the related category of industry.

13. A system comprising:

one or more processors configured to;

receive a set of search keywords input by a user;

determine whether the set of search keywords indicate a clearly defined intention by the user, including comparing the set of search keywords with a plurality of predetermined words corresponding to intentions that are not clearly defined, and comparing the set of search keywords with a plurality of predetermined words corresponding to intentions that are clearly defined; and

in the event that the set of search keywords are determined not to indicate a clearly defined intention, execute a first recommendation method to obtain recommended search keywords;

wherein;

the first recommendation method is selected among a plurality of recommendation methods; and

the first recommendation method includes a knowledge-based recommendation method, a session correlation-based method, or a combination; and

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

14. The system of claim 13, in the event that the set of search keywords are determined to indicate a clearly defined intention, the one or more processors are further configured to execute is a second recommendation method to obtain additional recommended search keywords.

15. The system of claim 14, wherein the second recommendation method includes a search log-based recommendation method.

16. The system of claim 13, wherein:

the first recommendation method includes a combination of at least a knowledge-based recommendation method and a session correlation-based recommendation method; and

the knowledge-based recommendation method contributes to a first proportion of recommended keywords and the session correlation-based recommendation method contributes to a second proportion of recommended keywords.

17. The system of claim 16, wherein the one or more processors are further configured to:

record a first exposure to feedback rate of the knowledge-based recommendation method and a second exposure to feedback rate of the session correlation-based recommendation method; and

adjust the first proportion of recommended keywords contributed by the knowledge-based recommendation method for future received search keyword sets, and the second proportion of recommended keywords contributed by the session correlation-based recommendation method for future received search keyword sets.

18. The system of claim 13, wherein the knowledge-based recommendation method comprises:

obtaining a plurality of lexical item sets based on search logs;

selecting a lexical item set from the plurality of lexical item sets, the selected lexical item set includes one or more lexical items that best match the set of search keywords;

selecting a selected rule from a stored rule set, wherein each rule in the stored rule set includes a priority-sequenced chain of lexical item types, and in the selected rule, the lexical item type of the selected lexical item set is the first item in the corresponding priority-sequenced chain;

identifying a second lexical item type in the priority-sequenced chain in the selected rule; and

determining a lexical item that is in the set of the second lexical item type and that is logically corresponds to the received search keyword set as one of the recommended search keywords.

19. The system of claim 18, wherein the plurality of lexical item sets include a compound-type lexical item set, a mono-type lexical item set, a brand-type lexical item set, and a product model-type lexical item set.

20. The system of claim 13, wherein the session correlation-based recommendation method comprises:

storing search keyword chains corresponding to search sessions by users;

identifying a plurality of search keyword sets listed after the received search keyword set in the stored search keyword chain comprising the received search keyword set; and

determining the recommended search keywords based on positions of the plurality of search keyword sets.

21. The system of claim 20, wherein determining the recommended search keywords includes:

based on pre-stored sets having lexical items of various categories of industries, selecting from the search keyword sets listed after the input search keywords in the search keyword chains a set of search keywords whose category of industry is the same as that of the received search keyword as recommended search keywords.

22. The system of claim 20, wherein determining the recommended search keywords includes:

calculating probabilities of different later-listed search keyword sets occurring among all later-listed search keyword sets; and

selecting, based on the probabilities, search keyword sets that are most likely to occur as recommended search keywords.

24. A computer program product for search keyword recommendation, the computer program product being embodied in a computer readable storage medium and comprising computer instructions for:

receiving a set of search keywords input by a user;

determining whether the set of search keywords indicate a clearly defined intention by the user, including comparing the set of search keywords with a plurality of predetermined words corresponding to intentions that are not clearly defined, and comparing the set of search keywords with a plurality of predetermined words corresponding to intentions that are clearly defined; and

in the event that the set of search keywords are determined not to indicate a clearly defined intention, executing a first recommendation method to obtain recommended search keywords;

wherein;

the first recommendation method is selected among a plurality of recommendation methods; and

the first recommendation method includes a knowledge-based recommendation method, a session correlation-based method, or a combination.