Natural Language Processing: What are the best algorithms, papers on entity extraction, relationship extraction from text?

Just like Yang Li said, the question asked is a typically entity linking problem, which tries to link the extracted mentions to the entities in the Wikipedia page. I am in a team which participate in this year's Entity Linking Task. Here are the papers you definitely should have looked at. 1. Local and Global Algorithms for Disambiguation to Wikipedia. UIUC Group In this paper, the authors try to form the problem as an optimization problem and provides a list of local and global features, which might be categorized as algorithm part. They also published their system, called http://cogcomp.cs.illinois.edu/page/software_view/Wikifier. You can use their system to do entity disambiguation. When it comes to entity extraction part, you should paper their system paper, called "GLOW TAC-KBP 2011 Entity Linking System". You can get to know how they extract the mentions from the query document. 2. A Neighborhood Relevance Model for Entity Linking, UMass Group As we know, context information is very important for entity linking. If you just focus on the query document, you just get little context for the extracted mentions. This way, the second paper tries to collect the global context for mentions across the entire corpus. 3. Ji Heng's several papers about their system for entity linking. You can go to her website(http://nlp.cs.rpi.edu/publication.html) and search entity linking. They also has existed software. You can get your hand dirty into this field by using their system. 4. HLTCOE Participation at TAC 2012: Entity Linking and Cold Start Knowledge Base Construction, JHU Group Structured prediction is a very hot topic in the field of machine learning. In this paper, they proposed that Structured Prediction Cascades algorithm can be used to deal with entity linking problem. However, according to their paper, they have not implemented it in their system. It is just an idea. So if you want to focus on the algorithm, this idea is deserved for a try. 5. Joint Coreference Resolution and Named-Entity Linking with Multi-pass Sieves, Washington Group Coreference resolution is a very close are with the entity linking. So in this paper, they try to combine the two questions together and the hypothesis is that both task can benefited from each other. There are also a lot of research going on in this field. Just stay tuned.

I believe you are looking for something a.k.a entity linking? There are a lot of publications recently on this topic. Just search "entity linking" or "entity disambiguation" on google. For off-the-shelf tools, you can refer to Wikifier from UIUC or AIDA from MPI.

Tutorials and papers: http://ejmeij.github.io/entity-linking-and-retrieval-tutorial/ or http://nlp.cs.rpi.edu/kbp/2015/elreading.html Existing systems: http://babelfy.org/index.jsp AIDA: https://gate.d5.mpi-inf.mpg.de/webaida/ Tagme (for queries): http://tagme.di.unipi.it/ Wikifier: http://cogcomp.cs.illinois.edu/page/demo_view/Wikifier