# A multi label text classification problem

I'm looking to solve a multi label text classification problem but I don't really know how to formulate it correctly so I can look it up.. Here is my problem :

Say I have the document "I want to learn NLP. I can do that by reading NLP books or watching tutorials on the internet. That would help me find a job in NLP."

I want to classify the sentences into 3 labels (for example) objective, method and result. The result would be :

objective : I want to learn NLP

method : I can do that by reading NLP books or watching tutorials on the internet.

result : That would help me find a job.


As you would have noticed, it's not a classical classification problem, since the classification here depends on the document structure (unless I'm wrong?)

Any idea of the key words to better describe the problem ? or how I might solve it ?

Many thanks!

• You don't have classes, and therefore you don't have classification. I think what you are looking for resembles Question Answering. Mar 11 at 19:49
• I was thinking about Q&A too.. but I thought that could be considered as a classification problem since the classes won't change : for each document, the task would be to classify each sentence as objective, method or result. Mar 11 at 19:58
• I'd say that the design depends on the details of the task, for example: do you always have exactly 3 sentences? Is the order always objective then method then result? Are there sentences which don't belong to any of the 3 classes? It might be a segmentation task for instance, or sequence labeling. Mar 12 at 18:07

Based on some discussions and on the commentaries, the conclusion is that this problem could be rather considered as one of the following NLP tasks (some of which are pretty similar..) :

1. Q&A (as suggested by @Akavall too)
2. Intent Classification (or NER)
3. One shot Learning
4. Semantic Role Labeling
5. Sequence Labeling (as suggested by @Erwan)

Thanks!