Live chat stream comment management

1. A method, comprising:

receiving new comments provided to a current session of a live chat stream;

analyzing content of the new comments;

for each of the new comments, adding, based on the analyzing, the new comment to one or more of a plurality of collections of comments;

identifying active collections of the plurality of collections, active collections including at least one new comment;

calculating, for each active collection, a first measure of semantic similarity between the collection and the remaining active collections of the live chat stream;

calculating, for each active collection, a second measure of semantic similarity between the active collection and non-active collections of the live chat stream;

calculating, for each active collection, a third measure of semantic similarity between the active collection and historical collections of comments for one or more prior sessions of the live chat stream;

determining, for each active collection, a semantic-uniqueness score based on the first, second and third measures of semantic similarity;

determining, for each active collection, an activity score representing an activity level of the active collection for the current session of the live chat stream;

determining, for each active collection, a final importance score based on its semantic-uniqueness score and activity score; and

controlling the display of each active collection based on its final importance score.

2. The method of claim 1 , wherein the determining a semantic-uniqueness score based on its first, second and third measures of semantic similarity includes applying weighting values to the first, second and third measures of semantic similarity.

3. The method of claim 1 , wherein the activity score is determined based on a temporally decaying value.

4. The method of claim 3 , wherein the determining an activity score for an active collection comprises:

increasing a previous activity score for the active collection responsive to a new comment being added to the active collection; and

decreasing a previous activity score for the active collection responsive to no new comment being added to the active collection within a pre-determined period of time.

5. The method of claim 1 , wherein analyzing content of a new comment provided to the current session of the live chat stream comprises:

identifying one or more concepts of the new comment;

determining a distance of the new comment from the active collections of comments; and

extracting one or more key words from the new comment.

6. The method of claim 5 , wherein identifying one or more concepts of the new comment includes processing the new comment with a concept mining algorithm.

7. The method of claim 5 , wherein the determining a distance of the new comment from the active collections of comments includes calculating a Levenshtein distance of the new comment from each of the active collections of comments.

8. The method of claim 5 , wherein the analyzing content of the new comment provided to the current session of the live chat stream further includes:

calculating, for each active collection, a similarity score of the new comment based on:

the identified one or more concepts of the new comment;

the determined distance of the new comment from the active collections of comments; and

one or more extracted keywords from the new comment; and

wherein adding the new comment to one or more of the collections of comments based on a result of the analysis includes adding, for each active collection, the new comment to the active collection if the calculated similarity score of the new comment for the active collection exceeds a threshold value.

9. The method of claim 1 , wherein the controlling the display of an active collection based on its final importance score includes displaying the active collection if its final importance score exceeds a predetermined value.

10. A system, comprising:

a memory; and

a processor coupled to the memory, the processor configured to:

receive new comments provided to a current session of a live chat stream;

analyze content of the new comments;

for each of the new comments, add, based on the analyzing, the new comment to one or more of a plurality of collections of comments;

identify active collections of the plurality of collections, active collections including at least one new comment;

calculate, for each active collection, a first measure of semantic similarity between the collection and the remaining active collections of the live chat stream;

calculate, for each active collection, a second measure of semantic similarity between the active collection and non-active collections of the live chat stream;

calculate, for each active collection, a third measure of semantic similarity between the active collection and historical collections of comments for one or more prior sessions of the live chat stream;

determine, for each active collection, a semantic-uniqueness score based on the first, second and third measures of semantic similarity;

determine, for each active collection, an activity score representing an activity level of the active collection for the current session of the live chat stream;

determine, for each active collection, a final importance score based on its semantic-uniqueness score and activity score; and

control the display of each active collection based on its final importance score.

11. The system of claim 10 , wherein the determining a semantic-uniqueness score based on its first, second and third measures of semantic similarity includes applying weighting values to the first, second and third measures of semantic similarity.

12. The system of claim 10 , wherein the activity score is determined based on a temporally decaying value.

13. The system of claim 12 , wherein the determining an activity score for an active collection comprises:

increasing a previous activity score for the active collection responsive to a new comment being added to the active collection; and

decreasing a previous activity score for the active collection responsive to no new comment being added to the active collection within a pre-determined period of time.

14. The system of claim 10 , wherein analyzing content of a new comment provided to the current session of the live chat stream comprises:

identifying one or more concepts of the new comment;

determining a distance of the new comment from the active collections of comments; and

extracting one or more key words from the new comment.

15. The system of claim 14 , wherein identifying one or more concepts of the new comment includes processing the new comment with a concept mining algorithm.

16. The system of claim 15 , wherein the determining a distance of the new comment from the active collections of comments includes calculating a Levenshtein distance of the new comment from each of the active collections of comments.

17. The system of claim 16 , wherein the analyzing content of the new comment provided to the current session of the live chat stream further includes:

calculating, for each active collection, a similarity score of the new comment based on:

the identified one or more concepts of the new comment;

the determined distance of the new comment from the active collections of comments; and

one or more extracted keywords from the new comment; and

wherein adding the new comment to one or more of the collections of comments based on a result of the analysis includes adding, for each active collection, the new comment to the active collection if the calculated similarity score of the new comment for the active collection exceeds a threshold value.

18. The system of claim 13 , wherein the controlling the display of an active collection based on its final importance score includes displaying the active collection if its final importance score exceeds a predetermined value.

19. A computer program product, the computer program product comprising a computer readable storage medium having program instructions embodied therewith, the program instructions executable by a computer to cause the computer to:

receive new comments provided to a current session of a live chat stream;

analyze content of the new comments;

for each of the new comments, add, based on the analyzing, the new comment to one or more of a plurality of collections of comments;

identify active collections of the plurality of collections, active collections including at least one new comment;

calculate, for each active collection, a first measure of semantic similarity between the collection and the remaining active collections of the live chat stream;

calculate, for each active collection, a second measure of semantic similarity between the active collection and non-active collections of the live chat stream;

calculate, for each active collection, a third measure of semantic similarity between the active collection and historical collections of comments for one or more prior sessions of the live chat stream;

determine, for each active collection, a semantic-uniqueness score based on the first, second and third measures of semantic similarity;

determine, for each active collection, an activity score representing an activity level of the active collection for the current session of the live chat stream;

determine, for each active collection, a final importance score based on its semantic-uniqueness score and activity score; and

control the display of each active collection based on its final importance score.

20. The computer program product of claim 19 , wherein the determining a semantic-uniqueness score based on its first, second and third measures of semantic similarity includes applying weighting values to the first, second and third measures of semantic similarity.