Ranking and marking messages in a group-based communication system using machine learning techniques
Methods, systems, apparatuses, devices, and computer program products are described. In a group-based communication system, a user may save posts for later (e.g., to reply to a message at a later time, to complete a task associated with a message at a later time). The system may use a machine learning model to determine to automatically mark a post for later for a user, for example, based on a set of features including at least a semantic embedding of the post. Additionally, or alternatively, the system may use a machine learning model to determine an order for displaying items (e.g., posts, reminders, files) within a user view (e.g., a later tab, a drafts tab, a threads tab, a files tab) for a user via a user interface. The system may update one or more machine learning models based on how users interact with the posts, user views, or both.
1 . A method for automatically marking messages in a group-based communication system for later, comprising:
receiving a new message posted to a group-based communication channel of the group-based communication system, wherein a first user is associated with the group-based communication channel and a second user authored the new message;
inputting one or more features associated with the new message, the group-based communication channel, the first user, the second user, or a combination thereof into a machine learning model, the one or more features comprising at least a semantic embedding of the new message;
determining, in response to the inputting the one or more features into the machine learning model, an output of the machine learning model indicating to mark the new message for later for the first user; and
sending, for display in a user interface of a user device operated by the first user, an indication marking the new message for later based at least in part on the output of the machine learning model.
2 . The method of claim 1 , wherein sending the indication marking the new message for later comprises:
sending, for display in the user interface of the user device, a suggestion to mark the new message for later;
receiving, via the user interface of the user device, an input approving the suggestion; and
adding the new message to a plurality of messages marked for later for the first user in response to the input approving the suggestion.
3 . The method of claim 2 , further comprising:
updating the machine learning model based at least in part on the input approving the suggestion.
4 . The method of claim 1 , further comprising:
sending, for display in the user interface of the user device, a suggestion to mark the new message for later;
receiving, via the user interface of the user device, an input declining the suggestion; and
refraining from adding the new message to a plurality of messages marked for later for the first user in response to the input declining the suggestion.
5 . The method of claim 4 , further comprising:
updating the machine learning model based at least in part on the input declining the suggestion.
6 . The method of claim 1 , wherein a third user is associated with the group-based communication channel, the method further comprising:
inputting one or more second features associated with the new message, the group-based communication channel, the third user, the second user, or a combination thereof into the machine learning model, the one or more second features comprising at least the semantic embedding of the new message; and
determining, in response to the inputting the one or more second features into the machine learning model, a second output of the machine learning model indicating not to mark the new message for later for the third user.
7 . The method of claim 1 , further comprising:
receiving, via a sidebar of the user interface of the user device, an input indicating selection of a later tab for the group-based communication system; and
sending, for display in a main pane of the user interface of the user device and in response to the input, a plurality of messages marked for later for the first user, the plurality of messages marked for later comprising at least the new message.
8 . The method of claim 1 , further comprising:
generating a feature of the one or more features indicating that the new message comprises text corresponding to a date, a time, or both, wherein the output of the machine learning model indicating to mark the new message for later for the first user is based at least in part on the new message comprising the text corresponding to the date, the time, or both.
9 . The method of claim 8 , further comprising:
automatically creating a reminder for the first user in the group-based communication system based at least in part on the new message comprising the text corresponding to the date, the time, or both, wherein a trigger time for the reminder is based at least in part on the date, the time, or both.
10 . The method of claim 1 , wherein the machine learning model is trained based at least in part on:
first data that includes prior messages marked for later by a plurality of users; and
second data that includes text data, user data, channel data, or any combination thereof associated with the prior messages marked for later, to learn relationships between the first data and the second data, such that the machine learning model is trained to output an indication of whether to mark the new message for later upon input of the one or more features.
11 . The method of claim 10 , wherein:
the machine learning model corresponds to a first group of the group-based communication system, the first group comprising at least the first user and the second user; and
a second machine learning model different from the machine learning model is trained, the second machine learning model corresponding to a second group of the group-based communication system, and the second machine learning model trained to output a second indication of whether to mark new messages for later for users of the second group.
12 . The method of claim 1 , further comprising:
deploying the machine learning model for the group-based communication system;
retrieving, from a plurality of messages associated with the first user in the group-based communication system, a first subset of the plurality of messages associated with the first user based at least in part on a filter criterion; and
determining, based at least in part on the first subset of the plurality of messages associated with the first user and the machine learning model, a second subset of the first subset of the plurality of messages to mark for later for the first user.
13 . The method of claim 1 , wherein the one or more features further comprise a first graphical embedding for the first user, a second graphical embedding for the second user, a third graphical embedding for the group-based communication channel, contents of the new message, relationship data for the first user and the second user, first interaction data for the first user with the group-based communication channel, second interaction data for a plurality of other users with the new message, historical data for the first user, group structure information for a group associated with the group-based communication channel, or any combination thereof.
14 . An apparatus for automatically marking messages in a group-based communication system for later, comprising:
a processor;
memory coupled with the processor; and
instructions stored in the memory and executable by the processor to cause the apparatus to:
receive a new message posted to a group-based communication channel of the group-based communication system, wherein a first user is associated with the group-based communication channel and a second user authored the new message;
input one or more features associated with the new message, the group-based communication channel, the first user, the second user, or a combination thereof into a machine learning model, the one or more features comprising at least a semantic embedding of the new message;
determine, in response to the inputting the one or more features into the machine learning model, an output of the machine learning model indicating to mark the new message for later for the first user; and
send, for display in a user interface of a user device operated by the first user, an indication marking the new message for later based at least in part on the output of the machine learning model.
15 . The apparatus of claim 14 , wherein the instructions to send the indication marking the new message for later are executable by the processor to cause the apparatus to:
send, for display in the user interface of the user device, a suggestion to mark the new message for later;
receive, via the user interface of the user device, an input approving the suggestion; and
add the new message to a plurality of messages marked for later for the first user in response to the input approving the suggestion.
16 . The apparatus of claim 14 , wherein the instructions are further executable by the processor to cause the apparatus to:
send, for display in the user interface of the user device, a suggestion to mark the new message for later;
receive, via the user interface of the user device, an input declining the suggestion; and
refrain from adding the new message to a plurality of messages marked for later for the first user in response to the input declining the suggestion.
17 . The apparatus of claim 14 , wherein a third user is associated with the group-based communication channel, and the instructions are further executable by the processor to cause the apparatus to:
input one or more second features associated with the new message, the group-based communication channel, the third user, the second user, or a combination thereof into the machine learning model, the one or more second features comprising at least the semantic embedding of the new message; and
determine, in response to the inputting the one or more second features into the machine learning model, a second output of the machine learning model indicating not to mark the new message for later for the third user.
18 . The apparatus of claim 14 , wherein the instructions are further executable by the processor to cause the apparatus to:
generate a feature of the one or more features indicating that the new message comprises text corresponding to a date, a time, or both, wherein the output of the machine learning model indicating to mark the new message for later for the first user is based at least in part on the new message comprising the text corresponding to the date, the time, or both.
19 . The apparatus of claim 18 , wherein the instructions are further executable by the processor to cause the apparatus to:
automatically create a reminder for the first user in the group-based communication system based at least in part on the new message comprising the text corresponding to the date, the time, or both, wherein a trigger time for the reminder is based at least in part on the date, the time, or both.
20 . A non-transitory computer-readable medium storing code for automatically marking messages in a group-based communication system for later, the code comprising instructions executable by a processor to:
receive a new message posted to a group-based communication channel of the group-based communication system, wherein a first user is associated with the group-based communication channel and a second user authored the new message;
input one or more features associated with the new message, the group-based communication channel, the first user, the second user, or a combination thereof into a machine learning model, the one or more features comprising at least a semantic embedding of the new message;
determine, in response to the inputting the one or more features into the machine learning model, an output of the machine learning model indicating to mark the new message for later for the first user; and
send, for display in a user interface of a user device operated by the first user, an indication marking the new message for later based at least in part on the output of the machine learning model.