this is for holding javascript data
AT deleted The_standard_NLP_pro.tex
almost 9 years ago
Commit id: bf1a5aea1e558431b3fdafc062457d44fd18ee7e
deletions | additions
diff --git a/The_standard_NLP_pro.tex b/The_standard_NLP_pro.tex
deleted file mode 100644
index bd6eb25..0000000
--- a/The_standard_NLP_pro.tex
+++ /dev/null
...
The standard NLP problem which is closest to answering yes/no questions
is the so-called \textit{Recognizing Text Entailment} (RTE) task.
In the standard formulation,
we have a paragraph of context and a phrase that either is or isn't
\textit{entailed} , i.e.\ can be decided to be true just based on the context.
The most prominent effort in this area is probably the
\textsc{Excitement EOP} academic project,%
\footnote{\url{http://hltfbk.github.io/Excitement-Open-Platform/}}
which is a full-fledged
RTE pipeline in Java that implements several algorithms
in a common framework.
Typical state-of-art RTE algorithms work on the principle of parse tree
alignment --- grammar dependency tree of the hypothesis and each context
sentence is compared and we try to learn which changes in the tree might
keep entailment.
But as the RTE problems are quite hard and practical applications are limited,
this is not a very lively area of research per se.
diff --git a/layout.md b/layout.md
index f45779b..cf300fd 100644
--- a/layout.md
+++ b/layout.md
...
When_considering_the.tex
subsectionVector.tex
subsectionNLP_for_Ye.tex
The_standard_NLP_pro.tex
With_this_in_mind.tex
beginitemize__item_t.tex
To_conclude_the_yesn.tex