Facilitating a meeting using graphical text analysis

• US 10,346,539 B2
• Filed: 12/19/2016
• Issued: 07/09/2019
• Est. Priority Date: 11/03/2014
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Cecchi, Guillermo A., Kozloski, James R., Pickover, Clifford A., Rish, Irina
• Primary Examiner(s): Jackson, Jakieda R
• Application Number:
  US15/383,321
• Publication Number:
  US 20170097929A1
• Time in Patent Office: 932 Days
• Field of Search: 704 9, 704245
• CPC Class Codes: G06F 16/35 Clustering; Classification
• CPC Class Codes: G06F 16/358 Browsing; Visualisation therefor
• CPC Class Codes: G06F 40/30 Semantic analysis
• CPC Class Codes: G06Q 10/109 Time management, e.g. calendars, reminders, meetings or time accounting
• CPC Class Codes: G09B 5/02 with visual presentation of the material to be studied, e.g. using film strip

Abstract

Embodiments relate to facilitating a meeting. A method for facilitating a meeting of a group of participants is provided. The method generates a graph of words from speeches of the participants as the words are received from the participants. The method partitions the group of participants into a plurality of subgroups of participants. The method performs a graphical text analysis on the graph to identify a cognitive state for each participant and a cognitive state for each subgroup of participants. The method informs at least one of the participants about the identified cognitive state of a participant or a subgroup of participants.

Claims

1. A computer-implemented method for conducting a digital meeting of a group of participants, the method comprising:

partitioning, by a processor, the group of participants into a first subgroup of participants and a second subgroup of participants;

generating, by the processor, a respective graph for each participant, each respective graph including a plurality of nodes and a plurality of edges, each node representing a word or a phrase uttered by a participant and an identity of the participant, and each edge representing a temporal precedence of utterance of the words;

for each participant;

extracting, by the processor, a first set of topological features from the respective graph for the participant, the first set of topological features comprising topological measures of a graph skeleton of the graph including at least one of degree distribution, density of small-size motifs, clustering, or centrality;

extracting, by the processor, a second set of topological features from the respective graph for the participant based at least in part on feature vectors associated with the plurality of nodes; and

combining, by the processor, the first set of topological features and the second set of topological features to generate a multi-dimensional feature vector that represents a speech sample of the participant, wherein the multi-dimensional feature vector incorporates syntactic, semantic, and dynamical components of the speech sample;

training, by the processor, a classifier using the multi-dimensional feature vector generated for each participant to discriminate speech samples that belong to different conditions;

determining, by the processor, a respective cognitive state of each participant using the trained classifier, wherein determining the respective cognitive state of a participant comprises;

generating, in a multi-dimensional text feature space, a first set of plots of feature vectors of previously generated graphs associated with known cognitive states to form clusters;

generating, in the multi-dimensional text feature space, a second set of plots of the feature vectors of the respective graph for the participant;

determining, based at least in part on a distance between the first set of plots and the second set of plots, that the feature vectors of the respective graph for the participant fall in a feature space of a particular cluster; and

determining that the respective state of the participant corresponds to a cognitive state represented by the particular cluster;

determining, by the processor, a first combined cognitive state of the first subgroup of participants based on the respective cognitive states of the one or more participants from the first subgroup, wherein determining the first combined cognitive state comprises;

retrieving, from an information repository, weights associated with respective participants of the first subgroup of participants, wherein a weight assigned to each respective participant is based on an assigned role of each respective participant that is automatically retrieved and analyzed from an information repository; and

combining, the cognitive states of the participants of the first subgroup according to the corresponding weights;

detecting, by the processor, a change in the first combined cognitive state of the first subgroup of participants; and

in response, informing electronically the second subgroup of participants of the change in the first combined cognitive state of the first subgroup of participants.

2. The computer implemented method of claim 1, wherein the change in the first cognitive state of the first subgroup is detected in response to identifying a change in a cognitive state of a participant from the first subgroup.

3. The computer implemented method of claim 1, wherein detecting the change in the first cognitive state of the first subgroup of participants further comprises predicting the change in the first cognitive state based on one or more prior graphs.

4. The computer implemented method of claim 3, wherein in response to a confidence value of a prediction in the change in the first cognitive state being above a predetermined threshold, the second subgroup of participants is informed of the change in the first cognitive state.

5. The computer implemented method of claim 3, wherein in response to a confidence value of a prediction in the change in the first cognitive state being below a predetermined threshold:

accessing, by the processor, an external information repository for information associated with a participant from the first subgroup; and

boosting, by the processor, the confidence value of the prediction based on the information from the external information repository.

6. The computer implemented method of claim 5, wherein the information from the external information repository includes a biometric information of the participant, social network content of the participant, and speeches made by the participant in a prior meeting.

7. The computer-implemented method of claim 1, further comprising:

obtaining, by the processor, physiological measurements of the participants; and

determining, by the processor, respective cognitive states of one or more participants from the first subgroup further based at least in part on the physiological measurements.

8. The computer-implemented method of claim 1, further comprising:

receiving, by the processor, speech data corresponding to utterances of one or more participants from one or more client devices; and

performing, by the processor, emotion recognition processing based at least in part on the speech data to identify, in real-time, features present in the speech data, wherein the features comprise at least one of emphasis, intent, attitude, or semantic meaning in the utterances of the one or more participants.

9. The computer-implemented method of claim 1, wherein the second subgroup of participants comprises an instructor presenting a lesson plan to the first subgroup of participants, the method further comprising presenting a recommendation to the instructor to change the lesson plan based at least in part on the change in the first combined cognitive state.

10. A system for a digital meeting of a group of participants, the system comprising:

a memory;

a network interface; and

a processor coupled with the memory and the network interface, the processor configured to;

partition the group of participants into a first subgroup of participants and a second subgroup of participants;

generate a respective graph for each participant, each respective graph including a plurality of nodes and a plurality of edges, each node representing a word or a phrase uttered by a participant and an identity of the participant, and each edge representing a temporal precedence of utterance of the words;

for each participant;

extract a first set of topological features from the respective graph for the participant, the first set of topological features comprising topological measures of a graph skeleton of the graph including at least one of degree distribution, density of small-size motifs, clustering, or centrality;

extract a second set of topological features from the respective graph for the participant based at least in part on feature vectors associated with the plurality of nodes; and

combine the first set of topological features and the second set of topological features to generate a multi-dimensional feature vector that represents a speech sample of the participant, wherein the multi-dimensional feature vector incorporates syntactic, semantic, and dynamical components of the speech sample;

train a classifier using the multi-dimensional feature vector generated for each participant to discriminate speech samples that belong to different conditions;

determine a respective cognitive state of each participant using the trained classifier, wherein determining the respective cognitive state of participant comprises;

generating, in a multi-dimensional text feature space, a first set of plots of feature vectors of previously generated graphs associated with known cognitive states to form clusters;

generating, in the multi-dimensional text feature space, a second set of plots of the feature vectors of the respective graph for the participant;

determining, based at least in part on a distance between the first set of plots and the second set of plots, that the feature vectors of the respective graph for the participant fall in a feature space of a particular cluster; and

determining that the respective state of the participant corresponds to a cognitive state represented by the particular cluster;

determine a first combined cognitive state of the first subgroup of participants based on the respective cognitive states of the one or more participants from the first subgroup, wherein determining the first combined cognitive state comprises;

retrieving, from an information repository, weights associated with respective participants of the first subgroup of participants, wherein a weight assigned to each respective participant is based on an assigned role of each respective participant that is automatically retrieved and analyzed from an information repository; and

combining, the cognitive states of the participants of the first subgroup according to the corresponding weights;

predict a change in the first combined cognitive state of the first subgroup of participants; and

in response, inform electronically the second subgroup of participants of the change in the first combined cognitive state of the first subgroup of participants.

11. The system of claim 10, wherein the change in the first cognitive state of the first subgroup is detected in response to identifying a change in a cognitive state of a participant from the first subgroup.

12. The system of claim 10, wherein the processor predicts the change in the first cognitive state of the first subgroup of participants based on one or more prior graphs.

13. The system of claim 12, wherein in response to a determination that a confidence value of a prediction in the change in the first cognitive state is above a predetermined threshold, the processor informs the second subgroup of participants of the change in the first cognitive state.

14. The system of claim 12, wherein in response to a confidence value of a prediction in the change in the first cognitive state being below a predetermined threshold, the processor is configured to:

access an external information repository for information associated with a participant from the first subgroup; and

update the confidence value of the prediction based on the information from the external information repository.

15. The system of claim 14, wherein the information from the external information repository includes a biometric information of the participant, social network content of the participant, and speeches made by the participant in a prior meeting.

16. The system of claim 10, wherein the first subgroup includes a teacher and the second subgroup includes students learning from the teacher in the digital meeting.

17. A computer program product, comprising a non-transitory computer readable storage medium having program instructions embodied therewith, the program instructions readable by a processing circuit to cause the processing circuit to perform a method of conducting a digital meeting of a group of participants, the computer readable storage medium comprising instructions to:

partition the group of participants into a first subgroup of participants and a second subgroup of participants;

generate a respective graph for each participant, each respective graph including a plurality of nodes and a plurality of edges, each node representing a word or a phrase uttered by a participant and an identity of the participant, and each edge representing a temporal precedence of utterance of the words;

for each participant;

extract a first set of topological features from the respective graph for the participant, the first set of topological features comprising topological measures of a graph skeleton of the graph including at least one of degree distribution, density of small-size motifs, clustering, or centrality;

extract a second set of topological features from the respective graph for the participant based at least in part on feature vectors associated with the plurality of nodes; and

combine the first set of topological features and the second set of topological features to generate a multi-dimensional feature vector that represents a speech sample of the participant, wherein the multi-dimensional feature vector incorporates syntactic, semantic, and dynamical components of the speech sample;

train a classifier using the multi-dimensional feature vector generated for each participant to discriminate speech samples that belong to different conditions;

determine a respective cognitive state of each participant using the trained classifier, wherein determining the respective cognitive state of participant comprises;

generating, in a multi-dimensional text feature space, a first set of plots of feature vectors of previously generated graphs associated with known cognitive states to form clusters;

generating, in the multi-dimensional text feature space, a second set of plots of the feature vectors of the respective graph for the participant;

determining, based at least in part on a distance between the first set of plots and the second set of plots, that the feature vectors of the respective graph for the participant fall in a feature space of a particular cluster; and

determining that the respective state of the participant corresponds to a cognitive state represented by the particular cluster;

determine a first combined cognitive state of the first subgroup of participants based on the respective cognitive states of the one or more participants from the first subgroup, wherein determining the first combined cognitive state comprises;

retrieving, from an information repository, weights associated with respective participants of the first subgroup of participants, wherein a weight assigned to each respective participant is based on an assigned role of each respective participant that is automatically retrieved and analyzed from an information repository; and

combining, the cognitive states of the participants of the first subgroup according to the corresponding weights;

predict a change in the first combined cognitive state of the first subgroup of participants; and

in response, inform electronically the second subgroup of participants of the change in the first combined cognitive state of the first subgroup of participants.

18. The computer program product of claim 17, wherein the change in the first cognitive state of the first subgroup is detected in response to identifying a change in a cognitive state of a participant from the first subgroup.

19. The computer program product of claim 17, wherein the second subgroup of participants is informed of the change in the first cognitive state in response to a determination that a confidence value of a prediction in the change in the first cognitive state is above a predetermined threshold.

20. The computer program product of claim 17, wherein the computer readable storage medium further includes instructions to:

in response to a confidence value of a prediction in the change in the first cognitive state being below a predetermined threshold;

access an external information repository for information associated with a participant from the first subgroup; and

update the confidence value of the prediction based on the information from the external information repository.