Determination of prominent phrases in multi-channel interactions by multi-feature evaluations

1. A method for determining prominent keyphrases in interactions, comprising:

receiving keyphrases of numerous interactions received via a plurality of channels that comprise apparatuses for conveying the interactions according to the media thereof;

depending on each channel from said plurality of channels from which each keyphrase was received, quantitatively evaluating each keyphrase by a plurality of different metrics thereby yielding a corresponding plurality of values representing different aspects of prominence, and linearly combining the plurality of values with a corresponding plurality of coefficients associated with each channel from said plurality of channels via which an interaction having each keyphrase was received, thereby providing a score for each keyphrase; and

based on a condition related to the scores of the keyphrases, selecting at least one keyphrase as an at least one prominent keyphrase,

wherein the method is at least partially performed by an at least one computerized apparatus configured for interfacing with the apparatuses of the plurality of the channels.

2. The method according to claim 1 , wherein the corresponding plurality of coefficients is based on quantitatively grading keyphrases of a corpus of numerous interactions received via said plurality of channels.

3. The method according to claim 2 , wherein the grading is based on a semantic significance of the keyphrases.

4. The method according to claim 2 , wherein the corresponding plurality of the coefficients is obtained by applying a ranking model on the graded keyphrases and the plurality of different metrics.

5. The method according to claim 4 , wherein the ranking model is a Rank-SVM.

6. The method according to claim 1 , wherein the condition for selecting the at least one keyphrase comprises at least one keyphrase having a highest score.

7. The method according to claim 1 , wherein the condition for selecting the at least one keyphrase comprises a plurality of keyphrases having scores higher than the scores of the rest of the keyphrases.

8. The method according to claim 1 , wherein the condition for selecting the at least one keyphrase comprises keyphrases having a score higher than a given threshold.

9. The method according to claim 1 , wherein the condition for selecting the at least one keyphrase comprises keyphrases having scores that are higher by a difference greater than a certain threshold relative to the other scores.

10. The method according to claim 1 , wherein the condition for selecting the at least one keyphrase comprises keyphrases having scores within a certain variance.

11. The method according to claim 1 , wherein configured for interfacing with the apparatuses of the plurality of the channels implies configured for indirectly interfacing with the apparatuses of the plurality of the channels via an interfacing device.

12. The method according to claim 1 , wherein configured for interfacing with the apparatuses of the plurality of the channels implies configured for directly interfacing with the apparatuses of the plurality of the channels.

13. The method according to claim 1 , wherein keyphrases of interactions are obtained in advance based on textual analysis of the contents of the interactions.

14. An apparatus for carrying out the method according to claim 1 .

15. The method according to claim 1 , further comprising:

annotating the keyphrases received from said plurality of channels;

mapping the annotated keyphrases with said plurality of different metrics;

training a ranking model based on the mapped keyphrases; and

acquiring weights for each channel in said plurality of channels by applying the ranking model.