Information processing device, information processing method, and information processing program

Abstract

This information processing device acquires, from a chat database, speech information indicating a user ID for a participating user in each channel of a chat, and the speech volume of each participating user for each channel. The information processing device uses the speech information to train a user speech volume prediction model that outputs a speech volume prediction result for each participating user upon input of the user ID of the participating user of the chat. The user speech volume prediction model comprises: a user ID vector converter that converts a user ID of a participating user to a vector; and a speech volume predictor that receives input of a vector of a user ID of a participating user and predicts the speech volume of the participating user. After training the user speech volume prediction model, the information processing device outputs, in the user speech volume prediction model, information in which the user ID has been converted to a vector using the user ID vector converter.

Claims

1. A chat information acquisition unit acquires chat information for each chat channel, which includes the user IDs of the participating users in the channel and the amount of chat activity of each participating user in that channel. A user speech volume prediction model that outputs predicted speech volume results for each participant in a chat channel, using the embedding layer of a machine learning model, is trained to output predicted speech volume results for each participant in the chat channel by using the speech information as training data, and a user ID vectorizer that uses the embedding layer of a machine learning model to convert each user ID of a participant in a chat channel into a vector into which the characteristics of the amount of speech each participant gives when they join each channel and outputs it, and a speech volume predictor that takes the information converted from each participant's user ID into input and outputs predicted speech volume results for each participant in the channel, and a learning unit that uses the speech information as training data to train the user speech volume prediction model to output predicted speech volume results for each participant in the chat channel, An information output unit that inputs the user ID to the user ID vectorizer in the user speech volume prediction model after training, and outputs information obtained by converting the user ID output from the user ID vectorizer into a vector. An information processing device characterized by comprising:

2. The aforementioned number of statements is the number of times the user has made a statement. The information processing apparatus according to feature 1.

3. The aforementioned speech volume predictor uses a machine learning model to output a prediction result of the speech volume of the participating users. The information processing apparatus according to feature 1.

4. An information processing method performed by an information processing device, A step of obtaining message information for each chat channel, which shows the user IDs of the users participating in the channel and the amount of messages each of the participating users has sent in that channel. The process involves training a user speech volume prediction model, which outputs predicted speech volume results for each participant in a chat channel, using the speech information as training data, with the user ID vectorizer which uses the embedding layer of a machine learning model to convert each user ID of a chat channel into a vector into which the characteristics of the amount of speech each participant gives when they join each channel into numerical form and embed, and the speech volume predictor which takes the converted user IDs of each participant as input and outputs predicted speech volume results for each participant in the channel, so that the user speech volume prediction model outputs predicted speech volume results for each participant in the chat channel, and using the speech information as training data. The process involves inputting the user ID into the user ID vectorizer in the user speech volume prediction model after training, and outputting information obtained by vectorizing the user ID output from the user ID vectorizer. An information processing method characterized by including

5. An information processing program for causing a computer to function as an information processing device according to any one of claims 1 to 3.

Images