- Causality related pages
- Judea Pearl's IJCAI 99 talk
- His IJCAI 99 paper
- Judea Pearl's home page
- His article with Stuart Russel on Bayes nets.
- Weka toolkit
- Java based ASP+ reasoning system (wrapped around the Smodels system); Windows based ASP+ reasoning system.
- iQA: The intelligent QA system (slides that explain the theory behind this system) (new module RTE style)
In the box that appears, you can type a text followed by a question:
* The text is "X and Y and ... and Z are in the room". X, Y, Z etc.
can be (1) proper names that LCC's tool (LFT) recognizes as such, or
(2) phrases such as "three students", "two kids", etc. In this case, the
numbers should be in letters (due to a LFT limitation), between one and
nine, and the class of people can be "people", "professors", "kids",
"students" ("children" is also supported by our system, but about three
days ago LFT's began translating incorrectly sentences containing that
word, when I noticed they added a SR to the translation).
* The question is of the form "How many people (students/professors/etc.)
are in the room?". Notice that, if there are no elements of that class
in the room, then you will get no answer, rather than "0" (this can be
changed easily, but I haven't had a chance to get to it yet).
When you click "Run", the web server retrieves the logic form from
LCC (this takes a while), translates it to A-Prolog, reasons about
it, and shows the result.
The system also supports some questions dynamic domains, such as "X
and Y and ... died [in the accident]. How many ... are dead?" and "X
and Y and ... entered the room. How many ... are in the room
[now]?". (The parts in brackets denote words that you may or may not
type.)
- Knowledge representation and Question Answering: A chapter in the handbook of Knowledge Representation.
- The LCC logic form related slides and documents: knowledge representation from text, semantic calculus, Polaris definitions.
- Jim Martin's Spring 2007 NLP class in U of Colorado: draft version of updated chapters of his book with Daniel Jurafsky.
- MIT 6.864 Advanced NLP Class (Fall 2006)
- Handbook of Knowledge Representation
- Building a large annotated corpus of English: the Penn treebank (lists the POS tageset used in LFT)
- Recognizing textual entailment challenge (RTE-3)
- Some RTE example: categorized, not categorized
- Wordnet