Research Fields in NLP

nlp.md

NOTE: Your research does not have to be categorized to these labels.

Morphorogy- or Word-Level Problems

• morphological analysis
• word representation learning
• word sense disambiguation
• lexical relation identification (e.g., hypernymy identification)
• unknown words / rare words

Sentence-Level Problems

• sentence structure (i.e., syntax parsing):
  • Part-of-Speech (POS) tagging
  • chunking / shallow syntactic parsing
  • phrase-structure grammar parsing / PCFG parsing / constituency parsing
  • dependency grammar parsing
  • combinatory categorial grammar （CCG） parsing
  • head-driven phrase structure grammar （HPSG） parsing
  • tree-adjoining grammar (TAG) parsing
  • supertagging (i.e., identification of lexical items before constructing a tree; c.f., POS tagging）
• sentence meaning:
  • frame-semantic parsing
  • semantic role labeling (SRL) / predicate-argument structure
  • abstract meaning representation (AMR)
  • sentiment analysis (e.g., Stanford Sentiment Treebank dataset)
  • textual semantic similarity （including factoid QA models using Q→A retrieval）
  • paraphrase identification
  • recognizing textual entailment (RTE) / natural language inference （NLI）
• others:
  • grammatical error correction

Document-Level Problems

• document structure (i.e., discourse parsing):
  • coherence identification
  • discourse relation classification / shallow discourse parsing
  • discourse constituency parsing
  • discourse dependency parsing
  • anaphoric resolution, coreference resolution
• document meaning:
  • sentiment analysis (e.g., IMDB dataset)
  • category classification
  • reading comprehension （including non-factoid QA）

Generation-Related Problems

• language modeling and search algorithms
• machine translation
• paraphrase generation
• sentence compression (i.e., sentence-level summarization)
• summarization
• dialog

Knowledge Base-Oriented Problems / Information Retrieval

• factoid question answering
• knowledge-base representation learning, link prediction
• named entity recognition (NER)
• relation extraction
• entity linking

Cross-Lingual Problems

• learning shared representations across multiple languages
• utilizing typological information for improving learning models
• etc.

Multimodal Problems

• visual captioning
• visual question answering (VQA)
• multimodal machine translation
• visual storytelling
• news-headline generation & thumbnailing (Nakai's work)
• visual dialog
• etc.