Information Extraction
Input: Sentences or documents
Understand different entities and other things, and change it to a form like table that is useful to people.
Output
Semantics
- First order logic
- Inference
- Semantic analysis
∀x,y:Mother(x,y)⟹Parent(x,y)
Reading comprehension, Text understanding
Read documents, and answer questions related to it(Q & A).
Word sense disambiguation
WSD is important to MT
When the word is ambiguous in one language, it may be translated differently in another language.
Named entity recognition
Semantic role labeling
Coreference resolution
Identify the expressions that refer to the same entity in the discourse and to link them together.
- Anaphoric: The entity is mentioned at first, and then the expression is used to refer back to an entity.(2nd sentence)
- Cataphoric: The pronoun or the reference is introduced first before the entity itself is introduced.(3rd sentence)
Other tasks
- Ellipsis, parallelism and underspecification: a certain word is missing because it’s implied and can be understood from the context.
Chen speaks Chinese. I don’t.