Natural Language Annotation for Machine Learning by James Pustejovsky and Amber Stubbs

Author:James Pustejovsky and Amber Stubbs
Language: eng
Format: mobi, epub, pdf
Tags: COMPUTERS / Natural Language Processing
Publisher: O’Reilly Media
Published: 2012-10-10T16:00:00+00:00

MaxEnt works by keeping the entropy at a maximum while remaining consistent with the partial information that we have available to us, that is, the evidence. We will define any real-valued function of the context and the class to be a feature, fi(b,a) — these label-feature combinations are often called joint-features (as in Natural Language Processing with Python).

When using MaxEnt, first we need to identify the set of feature functions that will be most useful in our classification task. For each of these, we measure the expected value over our training data, and this becomes the constraint for the model distribution as seen in the following equation (see Nigam et al. 1999 for a good discussion of this).

When the constraints are estimated in this manner, the distribution will always be of the form shown in the following equation:

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.