Method and device for processing an electronic document

A device for processing an electronic document, comprising:

• a processor;
• and
• a memory storing instructions that, when executed by the processor, cause the process to perform a method comprising:
• splitting the text into at least one sentence;
• and
• for each sentence:
• associating each word of the sentence with a word-vector;
• representing the sentence by a sentence-vector, wherein said representing comprises computing a weighted average of word-vectors associated with the sentence to obtain the sentence-vector;
• if the sentence-vector is associated with a tag in a data set of sentence-vectors associated with tags, obtaining the tag from the data set;
• if the sentence-vector is not associated with a tag in the data set, obtaining a tag for the sentence-vector using a classification algorithm;
• and
• processing the sentence if the tag obtained for the sentence is associated with a predetermined label,
• the method further comprising assigning a weight to the word-vector, the weight assigned to the word-vector being inversely proportional to the probability of finding the word associated to the word-vector in the text.

A method of processing an electronic document comprising text, the method comprising:

• splitting the text into at least one sentence;
• and
• for each sentence:
• associating each word of the sentence with a word-vector;
• representing the sentence by a sentence-vector, wherein said representing comprises computing a weighted average of word-vectors associated with the sentence to obtain the sentence-vector;
• if the sentence-vector is associated with a tag in a data set of sentence-vectors associated with tags, obtaining the tag from the data set;
• if the sentence-vector is not associated with a tag in the data set, obtaining a tag for the sentence-vector using a classification algorithm;
• and
• processing the sentence if the tag obtained for the sentence is associated with a predetermined label,
• the method comprising normalizing the sentence-vector, wherein said normalizing comprises projecting the sentence-vector on a precomputed sentence-vector and subtracting the result of said projection from said sentence-vector.