Question routing for user communities

• US 9,508,104 B2
• Filed: 09/20/2013
• Issued: 11/29/2016
• Est. Priority Date: 09/20/2013
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Pal, Aditya, Wang, Fei
• Primary Examiner(s): Morrison, Jay
• Assistant Examiner(s): Hoang, Ken
• Application Number:
  US14/032,264
• Publication Number:
  US 20150088906A1
• Time in Patent Office: 1,166 Days
• Field of Search: 707/E17.014, 707/723, 707/769, 707/709, 707/748, 707/749, 707/999.003, 707/999.005, 707/E17.008, 707/706, 707/724, 707/725, 707/732, 707/741, 707/758, 707/999.007, 707/E17.002, 707/E17.005, 707/E17.01, 707/E17.032, 707/E17.033, 707/E17.046, 707/E17.099, 707/E17.109, 707/728, 707/731, 707/737, 707/738, 707/754, 707/771, 707/772, 707/780, 707/794, 707/999.001, 707/999.004, 707/999.1, 706/12, 706/45, 706/46, 706/52, 709/204, 709/206, 709/217, 709/224, 709/218, 709/219, 709/223, 709/225, 709/226, 705/319, 705/304, 705/14.42, 705/14.71, 705/1.1, 705/3, 705/35, 705/37, 705/7.11, 705/7.14, 705/7.32, 705/14.12, 705/14.31, 705/14.41, 705/14.49, 705/14.53, 705/14.54, 705/14.73, 705/26.1, 705/26.61, 705/26.64, 705/26.7, 705/314, 705/347, 705/40, 705/412, 705/7.13, 705/7.29, 705/7.38
• US Class Current: 1/1
• CPC Class Codes: G06F 16/24578 using ranking
• CPC Class Codes: G06Q 50/01 Social networking
• CPC Class Codes: H04L 67/01 Protocols
• CPC Class Codes: H04L 67/02 based on web technology, e.g. hypertext transfer protocol [HTTP]
• CPC Class Codes: H04L 67/306 User profiles

Abstract

A computer-implemented method routes a current question to one or more of a plurality of online communities. A computer system can determine, for the current question presented by an asking user a plurality of question-to-question similarity values, a plurality of question-to-user similarity values and a plurality of question-to-community similarity values. The system can select one or more of the plurality of online communities based on the similarity values. The system can route the current question presented by the asking user to the selected one or more of the plurality of online communities.

Claims

1. A computer-implemented method comprising:

determining, for a current question presented by an asking user and for a plurality of existing questions within a plurality of online communities that each has a plurality of users contributing answers to the plurality of existing questions, a plurality of question-to-question similarity values;

determining, for the current question presented by the asking user and of the plurality of users contributing answers, a plurality of question-to-user similarity values, wherein the question-to-question similarity values include;

values determined by how many times a particular question is viewed by a corresponding online community, how many unique users answered the particular question, whether or not another asking user answered the particular question, whether or not another asking user answered the particular question, a word vector of the existing question, and a language analysis of the particular question in a corresponding online community;

determining, for the current question presented by the asking user and the plurality of online communities, a plurality of question-to-community similarity values, the plurality of question-to-community similarity values including values determined from a word vector of a community title and description;

selecting one or more of the plurality of online communities based on the plurality of question-to-question similarity values, the plurality of question-to-user similarity values and the plurality of question-to-community similarity values; and

routing the current question presented by the asking user to the selected one or more of the plurality of online communities.

2. The method of claim 1, wherein selecting one or more of the plurality of online communities includes:

ranking, for the plurality of online communities, a plurality of similarity values that include;

the question-to-question similarity values, the question-to-user similarity values and the question-to-community similarity values;

identifying, based on the ranking, a subset of similarity values from the plurality of similarity values; and

scoring online communities of the plurality of online communities based on how many similarity values in the subset correspond to a particular online community being scored; and

wherein the selecting one or more of the plurality of online communities is responsive to the scoring of the online communities.

3. The method of claim 1, wherein the question-to-question similarity values include values determined from at least one factor selected from the group consisting of:

how many user views for an existing question in a corresponding online community, how many unique user answered the existing question, whether or not another asking user answered the existing question, a word vector of the existing question, topic distribution for the existing question and a language analysis of the existing question in a corresponding online community.

4. The method of claim 1, wherein the question-to-user similarity values include values determined from at least one factor selected from the group consisting of:

demographics of users, communities that a user has created, communities that users are members of, a topical expertise of the user, an availability of the user, an inclination of another user to respond to the asking user, and a language analysis of answered questions in the plurality of online communities.

5. The method of claim 1, wherein the question-to-community similarity values further include values determined from at least one factor selected from the group consisting of:

a topical distribution of a community, a topical expertise of community members, an availability of community members, an inclination of community members to respond to the asking user and a language analysis of community data content.

6. The method of claim 2, further comprising

determining similarity values for the plurality of online communities, the similarity values for the plurality of online communities including a topic similarity between the plurality of online communities;

partitioning, prior to the ranking and based upon the topic similarity, the plurality of online communities into different clusters of online communities;

determining, by comparing the current question to the different clusters, a most topically appropriate cluster for the current question; and

selecting, based on the determining a most topically appropriate cluster, a first cluster of the different clusters of online communities, wherein the ranking only occurs within the first cluster of the different clusters of online communities.

7. The method of claim 1, wherein determining a plurality of question-to-user similarity values includes determining a likelihood that a particular user will respond to the current question based on a strength of a shortest path between two or more users in a question answer (QA) graph, the QA graph specifying which users have answer questions for other users.

8. The method of claim 7, wherein the QA graph is weighted according to how many questions each user has answered for other users.

9. The method of claim 1, wherein the question-to-user similarity values include values determined from a topical expertise of a first user that is calculated based upon how many questions were answered by the first user for a particular topic and how many questions were asked by the first user for the particular topic.

10. A computer-implemented method comprising:

determining, for a current question presented by an asking user and a plurality of existing questions within a plurality of online communities, a plurality of similarity values for each of the plurality of online communities, wherein each of the plurality of online communities corresponds to a distinct website portal, and wherein the determining includes determining a plurality of question-to-question similarity values, the question-to-question similarity values including;

values determined by how many times a particular question is viewed by a corresponding online community, how many unique users answered the particular question, whether or not another asking user answered the particular question, whether or not another asking user answered the particular question, a word vector of the existing question, and a language analysis of the particular question in a corresponding online community;

ranking the plurality of similarity values;

selecting, based on the ranking, a subset of similarity values from the plurality of similarity values;

scoring the online communities of the plurality of online communities based on how many similarity values in the subset that correspond to a particular online community being scored;

selecting one or more of the plurality of online communities based on the scoring; and

routing the current question presented by the asking user to the selected one or more of the plurality of online communities.

11. The method of claim 10, wherein determining a plurality of similarity values for each of the plurality of online communities includes determining, for the current question presented by the asking user and of a plurality of users contributing answers to questions, a plurality of question-to-user similarity values.

12. The method of claim 10, wherein determining a plurality of similarity values for each of the plurality of online communities includes determining, for the current question presented by the asking user and the plurality of online communities, a plurality of question-to-community similarity values.

13. The method of claim 10, wherein selecting a subset of similarity values from the plurality of similarity values includes comparing similarity values from the plurality of similarity values globally across the plurality of online communities and then computing the similarity values through a voting mechanism.

14. The method of claim 10, wherein selecting a subset of similarity values from the plurality of similarity values includes comparing similarity values from the plurality of similarity values locally for a first community of the plurality of online communities and determining a first quantity of top objects for the first community, and wherein the selecting further includes comparing similarity values from a plurality of similarity values locally for a second community and determining a second quantity of top objects for the second community based on a similarity between the second community and the current question.

15. The method of claim 14, wherein selecting a subset of similarity values from the plurality of similarity values further includes comparing similarity values from the plurality of similarity values globally across the plurality of online communities.

17. The method of claim 11, wherein the question-to-user similarity values include:

communities that a user has created, an availability of the user, an inclination of another user to respond to an asking user, and a language analysis of answered question in the plurality of online communities.

18. The method of claim 12, wherein the question-to-community similarity values include:

a word vector of a community title and description, a topical distribution of a community, an availability of community members, and an inclination of community members to respond to the asking user and a language analysis of community data content.

16. A computer program product for routing questions, the computer program product comprising a non-transitory computer readable storage medium having program code embodied therewith, the program code readable/executable by a computer to:

determine, using the computer and for a current question presented by an asking user and for a plurality of existing questions within a plurality of online communities that each has a plurality of users contributing answers to the plurality of existing questions, a plurality of question-to-question similarity values, wherein the question-to-question similarity values include;

values determined by how many times a particular question is viewed by a corresponding online community, how many unique users answered the particular question, whether or not another asking user answered the particular question, whether or not another asking user answered the particular question, a word vector of the existing question, and a language analysis of the particular question in a corresponding online community

determine, using the computer and for the current question presented by the asking user and of the plurality of users contributing answers, a plurality of question-to-user similarity values;

determine, using the computer and for the current question presented by the asking user and the plurality of online communities, a plurality of question-to-community similarity values;

select, using the computer, one or more of the plurality of online communities based on the plurality of question-to-question similarity values, the plurality of question-to-user similarity values and plurality of question-to-community similarity values; and

route, using the computer, the current question presented by the asking user to the selected one or more of the plurality of online communities.