Terry

refactoring to work with the annotated plain text

1

sometime the training data maybe plain text, instead of using python-crfsuite or any other CRF package, i still prefer to use webstruct because it has sklearn `pipeline` and some evaluation...

allow assign the BIO tags with fuzzy match. this can be useful when having some of partially labelled dataset.

i.e. ``` python parent = elem.getparent() pos = parent.index(elem) ``` this should capture the case where first element is the subject and second is the content

Hi, thanks for the work on the thriftpy, it made us much easier to wrap up our internal machine learning service with it! however i wonder if it's possible to...

add MdrExtractor to parse the listing data. the output will be a separated field with the name as the group name set in the annotation (using listingDataGroupName) and the value...

memory issue for some urls

3

hey, we found following code will make urlclustering use lots of memory and ends up got killed by system. could you take a look? thanks ```python import urlclustering lines=""" http://www.digitecno.it/webconference.aspx...

[feature request] allow load the user customized alert class

1

user may develop new Alert class or customize existing one, would be nice to be able load these classes.