def view_from_jupyter(
self,
document_id: str,
metric: str or Callable = 'jensenshannon',
num_top_similar: int = 5,
num_digits: int = 3,
keep_similar_by_words: bool = True,
display_output: bool = True,
give_html: bool = False,):
"""
Method for viewing documents similar to requested one
from jupyter notebook. Provides document titles and snippets of
first few sentences.
Parameters
----------
document_id
ID of the document in `dataset`
metric
Distance measure which is to be used to measure how documents differ from each other
If str -- should be one of 'jensenshannon', 'euclidean', 'cosine', 'correlation' --
as in scipy.spatial.distance.cdist
If callable -- should map two vectors to numeric value
num_top_similar
How many top similar documents' IDs to show
keep_similar_by_words
Whether or not to keep in the output those documents
that are similar to the given one by their constituent words and words' frequencies
display_output
if provide output at the end of method run
give_html
return html string generated by the method
Returns
-------
topic_html
html string of the generated output
"""
from IPython.display import display_html
from topicnet.cooking_machine.pretty_output import make_notebook_pretty
make_notebook_pretty()
search_ids, search_distances = self.view(
document_id=document_id,
metric=metric,
num_top_similar=num_top_similar,
keep_similar_by_words=keep_similar_by_words,
)
topic_html = ''
search_ids = [document_id] + search_ids
search_distances = [0] + search_distances
for doc_id, distance in zip(search_ids, search_distances):
document = self._dataset.get_source_document(doc_id)
topic_html += prepare_doc_html_with_similarity(document, distance)
if display_output:
display_html(topic_html, raw=True)
if give_html:
return topic_html
def view_from_jupyter(
self,
current_num_top_doc: int = None,
topic_names: list = None,
display_output: bool = True,
give_html: bool = False,
):
"""
TopDocumentsViewer method recommended for use
from jupyter notebooks
Returns texts of the actual documents.
Parameters
----------
current_num_top_doc
number of top documents to provide for
each cluster (Default value = None)
topic_names
list of topic names to view
display_output
if provide output at the end of method run
give_html
return html string generated by the method
Returns
-------
html_output
html string of the output
"""
from IPython.display import display_html
from topicnet.cooking_machine.pretty_output import make_notebook_pretty
make_notebook_pretty()
html_output = []
doc_list = self.view(current_num_top_doc, topic_names=topic_names)
for topic_name, topic_docs_dict in doc_list.items():
topic_docs = list(topic_docs_dict.keys())
topic_html = ''
topic_headline = f'<h1><b>Topic name:</b> {topic_name}</h1>'
topic_html += topic_headline
for doc_id in topic_docs:
document = self._dataset.get_source_document(doc_id)
topic_html += prepare_html_string(document)
html_output.append(topic_html)
if display_output:
display_html('<br />'.join(html_output), raw=True)
if give_html:
return html_output
def view_from_jupyter(self,
display_output: bool = True,
give_html: bool = False,
**kwargs):
"""
TopicMapViewer method recommended for use
from jupyter notebooks
returns closest pairs of models topics
and visualizes their top tokens
The class provide information about top tokens
of the model topics providing with different methods to score that.
Parameters
----------
display_output
if provide output at the end of method run
give_html
return html string generated by the method
Returns
-------
out_html
html string of the output
Another Parameters
------------------
**kwargs
*kwargs* are optional `~.TopTokenViewer` properties
"""
from IPython.display import display_html
from topicnet.cooking_machine.pretty_output import make_notebook_pretty
if 'digits' in kwargs:
digits = kwargs.pop('digits')
else:
digits = 5
make_notebook_pretty()
first_model_order, second_model_order = self.view()
token_view = (TopTokensViewer(model=self.model,
**kwargs).view_from_jupyter(
topic_names=first_model_order,
digits=digits,
display_output=False,
give_html=True))
second_token_view = (TopTokensViewer(
model=self.second_model,
**kwargs).view_from_jupyter(topic_names=second_model_order,
digits=digits,
display_output=False,
give_html=True))
model_name = self.model.model_id
second_model_name = self.second_model.model_id
out_html = '<table style=display:inline; cellpadding="5";><tbody>{0}</tbody></table>'
first_element = (f'<tr><td> First model name: '
f'{model_name}</td><td> Second model '
f'name: {second_model_name}</td></tr>{{0}}')
out_html = out_html.format(first_element)
table_contents = []
for t1, t2 in zip(token_view, second_token_view):
table_contents += [f'<tr><td>{t1}</td><td>{t2}</td></tr>']
out_html = out_html.format(''.join(table_contents))
if display_output:
display_html(out_html, raw=True)
if give_html:
return out_html
def view_from_jupyter(
self,
topic_names: Union[str, List[str]] = None,
digits: int = 5,
horizontally_stack: bool = True,
display_output: bool = True,
give_html: bool = False,
):
"""
TopTokensViewer method recommended for use
from jupyter notebooks
Parameters
----------
topic_names
topics requested for viewing
digits
Number of digits to round each probability to
horizontally_stack
if True, then tokens for each modality will be stacked horizontally
(instead of being a single long multi-line DataFrame)
display_output
request for function to output the information
together with iterable output intended to be used
as user defined output
give_html
return html string generated by the method
Returns
-------
topic_html_strings: list of strings in HTML format
Examples
--------
>>> # model training here
>>> # ...
>>> viewer = TopTokensViewer(model)
>>> information = viewer.view_from_jupyter()
>>> # or
>>> information = viewer.view_from_jupyter(output=False)
"""
from IPython.core.display import display_html
from topicnet.cooking_machine.pretty_output import make_notebook_pretty
make_notebook_pretty()
if isinstance(topic_names, list):
pass
elif isinstance(topic_names, str):
topic_names = [topic_names]
elif topic_names is None:
topic_names = self._model.topic_names
else:
raise TypeError(f'Invalid type `topic_names` type: "{type(topic_names)}"')
topic_html_strings = []
for topic in topic_names:
topic_html = self.to_html(
topic_names=topic,
digits=digits,
horizontally_stack=horizontally_stack,
)
if display_output:
display_html(topic_html, raw=True)
topic_html_strings.append(topic_html)
if give_html:
return topic_html_strings