Generative system for writing entity recommendations
Methods, systems, and apparatuses include receiving input of a selection from a client device providing a graphical user interface and a recommendation interface, where the input of the selection includes an explicit attribute. Implicit attribute suggestions are generated based on the explicit attribute. The implicit attribute suggestions are sent to the client device. Attribute selections including at least one implicit attribute suggestion are received from the client device. Prompts are created based on the attribute selections. A generative language model is applied to the prompts. Content including a suggested user recommendation for the first user profile is output by the generative language model based on the prompts. The suggested user recommendation is sent to the client device to cause the suggested user recommendation to be presented on the recommendation interface.
1 . A method for aiding a second user to write a recommendation for a first user, comprising:
receiving, from a client device, an input comprising a selection of the second user, wherein the client device provides (i) a graphical user interface associated with the first user of a first user profile, presented to the second user and (ii) a recommendation interface presented to the second user, and the selection of the second user comprises at least one selected explicit attribute corresponding to the first user and included in the first user profile;
generating one or more implicit attribute suggestions corresponding to the first user, based on the at least one selected explicit attribute;
sending the one or more implicit attribute suggestions to the client device to be presented to the second user on the recommendation interface;
receiving, from the client device, one or more attribute selections comprising at least one implicit attribute suggestion corresponding to the first user, wherein the one or more attribute selections are generated by an interaction with the second user about the first user, at the recommendation interface, with the one or more implicit attribute suggestions;
creating one or more prompts based on the one or more received attribute selections;
applying a generative language model to the one or more prompts; and
outputting, by the generative language model, based on the one or more prompts, content comprising a suggested user recommendation for the first user profile, wherein the suggested user recommendation is based on the interaction with the second user and is capable of being added to the first user profile.
2 . The method of claim 1 , further comprising:
generating one or more explicit attribute suggestions based on the at least one explicit attribute; and
sending the one or more explicit attribute suggestions to the client device to be presented on the recommendation interface, wherein the one or more received attribute selections are further generated by an interaction, at the recommendation interface, with the one or more explicit attribute suggestions.
3 . The method of claim 2 , further comprising:
generating a qualifier for each of the one or more explicit attribute suggestions; and
sending, to the client device, the qualifier with the one or more explicit attribute suggestions.
4 . The method of claim 1 , wherein generating one or more implicit attribute suggestions comprises:
applying a language model to the at least one explicit attribute, wherein the language model outputs the one or more implicit attribute suggestions.
5 . The method of claim 4 , further comprising:
classifying, by the language model, the one or more implicit attribute suggestions as implicit.
6 . The method of claim 1 , wherein generating one or more implicit attribute suggestions comprises:
filtering out attribute suggestions of the one or more implicit attribute suggestions based on similarity between attribute suggestions.
7 . The method of claim 1 , further comprising:
receiving, from the client device, feedback;
creating one or more updated prompts based on one or more received attribute selections and the feedback;
applying the generative language model to the one or more updated prompts;
outputting, by the generative language model, based on the one or more updated prompts, an updated suggested user recommendation; and
sending, to the client device, the updated suggested user recommendation.
8 . The method of claim 1 , further comprising:
receiving, from the client device, a recommendation submission; and
sending the suggested user recommendation output by the generative language model to the client device to be presented on the recommendation interface in response to receiving the recommendation submission.
9 . The method of claim 1 , further comprising:
receiving, from the client device, user input, wherein creating the one or more prompts is further based on the user input.
10 . The method of claim 1 , further comprising:
determining a second user profile, wherein the second user profile is associated with the second user of the graphical user interface, wherein creating the one or more prompts is further based on the first user profile and the second user profile.
11 . A system for aiding a second user to write a recommendation for a first user comprising:
at least one memory device; and
a processing device, operatively coupled with the at least one Memory device, to:
receive, from a client device, an input comprising a selection of the second user, wherein the client device provides (i) a graphical user interface associated with the first user of a first user profile, presented to the second user and (ii) a recommendation interface presented to the second user, and the selection of the second user comprises at least one selected explicit attribute corresponding to the first user and included in the first user profile;
generate one or more implicit attribute suggestions corresponding to the first user based on the at least one selected explicit attribute;
send the one or more implicit attribute suggestions to the client device to be presented to the second user on the recommendation interface;
receive, from the client device, one or more attribute selections comprising at least one implicit attribute suggestion corresponding to the first user, wherein the one or more attribute selections are generated by an interaction with the second user about the first user, at the recommendation interface, with the one or more implicit attribute suggestions;
create one or more prompts based on the one or more received attribute selections;
apply a generative language model to the one or more prompts; and
output, by the generative language model, based on the one or more prompts, content comprising a suggested user recommendation for the first user profile, wherein the suggested user recommendation is based on the interaction with the second user and is capable of being added to the first user profile.
12 . The system of claim 11 , wherein the processing device is further to:
generate one or more explicit attribute suggestions based on the at least one explicit attribute; and
send the one or more explicit attribute suggestions to the client device to be presented on the recommendation interface, wherein the one or more received attribute selections are further generated by an interaction, at the recommendation interface, with the one or more explicit attribute suggestions.
13 . The system of claim 12 , wherein the processing device is further to:
generate a qualifier for each of the one or more explicit attribute suggestions; and
send, to the client device, the qualifier with the one or more explicit attribute suggestions.
14 . The system of claim 11 , wherein generating one or more implicit attribute suggestions comprises:
applying a language model to the at least one explicit attribute, wherein the language model outputs the one or more implicit attribute suggestions.
15 . The system of claim 14 , wherein the processing device is further to:
classify, by the language model, the one or more implicit attribute suggestions as implicit.
16 . The system of claim 11 , wherein generating one or more implicit attribute suggestions comprises:
filtering out attribute suggestions of the one or more implicit attribute suggestions based on similarity between attribute suggestions.
17 . A non-transitory computer-readable storage medium for aiding a second user to write a recommendation for a first user comprising instructions that, when executed by a processing device, cause the processing device to:
receive, from a client device, an input comprising a selection of the second user, wherein the client device provides (i) a graphical user interface associated with the first user of a first user profile, presented to the second user and (ii) a recommendation interface presented to the second user, and the selection of the second user comprises at least one explicit attribute corresponding to the first user included in the first user profile;
generate one or more implicit attribute suggestions corresponding to the first user based on the at least one explicit attribute;
send the one or more implicit attribute suggestions to the client device to be presented to the second user on the recommendation interface;
receive, from the client device, one or more attribute selections comprising at least one implicit attribute suggestion corresponding to the first user, wherein the one or more attribute selections are generated by an interaction with the second user, at the recommendation interface, with the one or more implicit attribute suggestions;
create one or more prompts based on the one or more received attribute selections;
apply a generative language model to the one or more prompts; and
output, by the generative language model, based on the one or more prompts, content comprising a suggested user recommendation for the first user profile, wherein the suggested user recommendation is based on the interaction with the second user, at the recommendation interface, with the one or more implicit attribute suggestions and is capable of being added to the first user profile.
18 . The non-transitory computer-readable storage medium of claim 17 , wherein the processing device is further to:
generate one or more explicit attribute suggestions based on the at least one explicit attribute; and
send the one or more explicit attribute suggestions to the client device to be presented on the recommendation interface, wherein the one or more received attribute selections are further generated by an interaction, at the recommendation interface, with the one or more explicit attribute suggestions.
19 . The non-transitory computer-readable storage medium of claim 18 , wherein the processing device is further to:
generate a qualifier for each of the one or more explicit attribute suggestions; and
send, to the client device, the qualifier with the one or more explicit attribute suggestions.
20 . The non-transitory computer-readable storage medium of claim 17 , wherein generating one or more implicit attribute suggestions comprises:
applying a language model to the at least one explicit attribute, wherein the language model outputs the one or more implicit attribute suggestions.