NLP Binary Classification Task (using BERT)
Budget: $30 – $250 USD
trying to do a binary classification of whether or not a summary was created from a particular document.
(Let the summary be Si, and each sentence of the summary Si be Sij, and the document be Dk, and each sentence of the document Dk be Dkl.)
The summary Si is about 3 to 10 sentences, and a particular document Dk is more than 1000 sentences.
<The characteristics of the dataset used are as follows.>
There is a label to indicate whether or not a summary Si was created from a particular document Dk.
There is no label to indicate which sentence of the summary Sij was created from which sentence of the specific document Dkl.
There is no label to indicate which sentence of the summary Sij is made from which sentence of the specific document Dkl.
Since there are many sentences of similar topics, one or two sentences of the summary Sij may be made from different documents Dk.
In many cases, keywords are used as superordinate concepts in the summary Sij. (For example, superordinating car to vehicle.)
(Let the summary be Si, and each sentence of the summary Si be Sij, and the document be Dk, and each sentence of the document Dk be Dkl.)
The summary Si is about 3 to 10 sentences, and a particular document Dk is more than 1000 sentences.
<The characteristics of the dataset used are as follows.>
There is a label to indicate whether or not a summary Si was created from a particular document Dk.
There is no label to indicate which sentence of the summary Sij was created from which sentence of the specific document Dkl.
There is no label to indicate which sentence of the summary Sij is made from which sentence of the specific document Dkl.
Since there are many sentences of similar topics, one or two sentences of the summary Sij may be made from different documents Dk.
In many cases, keywords are used as superordinate concepts in the summary Sij. (For example, superordinating car to vehicle.)