Automated content recommendation in conferencing

1 . A method, comprising:

receiving a plurality of utterances associated with a plurality of participants of a communication session in real time during the communication session;

for each utterance, determining, by one or more pre-trained language learning models, whether a prediction of relatedness is present between the utterance and one or more trigger phrases associated with a content recommendation action from a list of content recommendation actions;

determining that a prediction of relatedness is present based on a semantic similarity score based on a cosine similarity between an embedding of the utterance and an embedding of the one or more trigger phrases exceeding a similarity threshold; and

transmitting, to one or more client devices, one or more pieces of content for display on the one or more client devices.

2 . The method of claim 1 , further comprising:

receiving a plurality of pieces of content; and

extracting, from each piece of content, a plurality of keywords representing the content;

for each utterance, determining whether a prediction of relatedness is present between the utterance and the plurality of keywords representing the content; and

transmitting, to one or more client devices, one or more pieces of content from which a related keyword was extracted.

3 . The method of claim 1 , wherein a party associated with each trigger phrase may be one or both of: a recipient party, and a customer party.

4 . The method of claim 1 , wherein receiving the list of content recommendation actions comprises:

presenting, to a client device associated with a user, a user interface (UI) comprising a prompt for the user to submit one or more content recommendation actions and one or more trigger phrases associated with the content recommendation actions.

5 . The method of claim 1 , wherein at least a subset of the content recommendation actions comprises at least one of: a recommended response to be uttered, one or more pieces of content to be recommended, and one or more content links to be provided.

6 . The method of claim 1 , wherein the one or more pieces of content to be recommended are directed to a recipient party to provide a response to a customer party.

7 . The method of claim 6 , wherein the recipient party comprises one or more of: a sales agent, a customer service agent, and a technical support agent.

8 . The method of claim 1 , wherein determining whether the predictions of relatedness are present further comprises determining whether a prediction of relatedness is present between the utterance and one or more variations on the one or more trigger phrases associated with the content recommendation action.

9 . The method of claim 1 , wherein determining whether a prediction of relatedness is present is performed at least in part by one or more sentence embedding models.

10 . The method of claim 1 , further comprising:

generating, based on the received list of content recommendation actions, one or more additional trigger phrases to be associated with one or more of the content recommendation actions.

11 . The method of claim 1 , further comprising:

detecting that one of the trigger phrases has been associated with a content recommendation action that differs in intent from the trigger phrase; and

associating the trigger phrase with a different content recommendation action.

12 . The method of claim 1 , wherein a subset of the trigger phrases are each associated with a plurality of content recommendation actions.

13 . A communication system comprising:

one or more processors configured to:

receive a plurality of utterances associated with a plurality of participants of a communication session in real time during the communication session;

for each utterance, determine, by one or more pre-trained language learning models, whether a prediction of relatedness is present between the utterance and one or more trigger phrases associated with a content recommendation action from a list of content recommendation actions;

determine that a prediction of relatedness is present based on a semantic similarity score based on a cosine similarity between an embedding of the utterance and an embedding of the one or more trigger phrases exceeding a similarity threshold; and

transmit, to one or more client devices, one or more pieces of content for display on the one or more client devices.

14 . The communication system of claim 13 , wherein the one or more processors are configured to determining whether the prediction of relatedness is present at least in part by one or more intent detection algorithms.

15 . The communications system of claim 13 , wherein the one or more pieces of content to be recommended are directed to a recipient party to provide a response to a customer party.

16 . The communications system of claim 13 , wherein determining whether the predictions of relatedness are present further comprises determining whether a prediction of relatedness is present between the utterance and one or more variations on the one or more trigger phrases associated with the content recommendation action.

17 . A non-transitory computer-readable medium comprising instructions, that when executed by one or more processors, causes the one or more processors to perform operations comprising:

receiving a plurality of utterances associated with a plurality of participants of a communication session in real time during the communication session;

for each utterance, determining, by one or more pre-trained language learning models, whether a prediction of relatedness is present between the utterance and one or more trigger phrases associated with a content recommendation action from a list of content recommendation actions;

determining that a prediction of relatedness is present based on a semantic similarity score based on a cosine similarity between an embedding of the utterance and an embedding of the one or more trigger phrases exceeding a similarity threshold; and

transmitting, to one or more client devices, one or more pieces of content for display on the one or more client devices.

18 . The non-transitory computer-readable medium of claim 17 , wherein a party associated with each trigger phrase may be one or both of: a recipient party, and a customer party.

19 . The non-transitory computer-readable medium of claim 17 , wherein the operations further comprise:

presenting, to a client device associated with a user, a user interface (UI) comprising a prompt for the user to submit one or more content recommendation actions and one or more trigger phrases associated with the content recommendation actions.

20 . The non-transitory computer-readable medium of claim 17 , wherein at least a subset of the content recommendation actions comprises at least one of: a recommended response to be uttered, one or more pieces of content to be recommended, and one or more content links to be provided.