Computer-implemented method and device for processing data

A computer-implemented method for machine learning, the method comprising:

• training a model for automatically classifying text data, the model being trained as a function of a representation of the text data, the text data including a plurality of words, the representation of the text data including, for each respective word of the plurality of words, a total word vector, the total word vector being generated by generating at least two word vectors for the respective word, calculating a weight for each of the at least two word vectors, and generating the total word vector as a weighted combination of the at least two word vectors according to the calculated weights, each calculated weight being a function of at least one property of the respective word in the text data;
• using, by a processor, the model to predict a classification of text data to be processed;
• generating, by the processor, a database based on the classification;
• and
• outputting, at an interface connected by at least one data line to the processor, the database.

A device for processing text data, the device comprising:

• a processor;
• and
• a memory for an artificial neural network;
• and
• an interface connected to the processor by at least one data line;
• the device configured to:
• receive, at the interface, the text data;
• generate, by the processor, a representation of the text data, the text data including a plurality of words, the generating including, for each respective word of the plurality of words:
• generating at least two word vectors, calculating a weight for each of the at least two word vectors, and generating a total word vector as a weighted combination of the at least two word vectors according to the calculated weights, each calculated weight being a function of at least one property of the respective word in the text data;
• use, by the processor, a model to predict a classification of the text data as a function of the representation of the text data;
• generate, by the processor, the database based on the classification of the text data;
• and
• output, at the interface, the database.

A device for machine learning, the device comprising:

• a processor;
• and
• a memory for an artificial neural network;
• and
• an interface connected to the processor by at least one data line;
• wherein the device is configured to:
• train a model for automatically classifying text data, the model being trained as a function of a representation of the text data, the text data including a plurality of words, the representation of the text data including, for each respective word of the plurality of words, a total word vector, the total word vector being generated by generating at least two word vectors for the respective word, calculating a weight for each of the at least two word vectors, and generating the total word vector as a weighted combination of the at least two word vectors according to the calculated weights, each calculated weight being a function of at least one property of the respective word in the text data;
• use, by the processor, the model to predict a classification of text data to be processed;
• generate, by the processor, a database based on the classification;
• and
• output, at the interface, the database.

A non-transitory computer-readable memory medium on which is stored a computer program, the computer program including computer-readable instructions for processing text data, for automatically extracting pieces of information from the text data and/or in models for creating a database, the instructions, when executed by a computer, causing the computer to perform the following steps:

• receiving, at an interface, the text data, the interface connected to a processor by at least one data line;
• generating, by the processor, a representation of the text data, the text data including a plurality of words, the generating including, for each respective word of the plurality of words:
• generating at least two word vectors, calculating a weight for each of the at least two word vectors, and generating a total word vector as a weighted combination of the at least two word vectors according to the calculated weights, each calculated weight being a function of at least one property of the respective word in the text data;
• using, by the processor, a model to predict a classification of the text data as a function of the representation of the text data;
• generating, by the processor, the database based on the classification of the text data;
• and
• outputting, at the interface, the database.