Update of install instructions.

README.md

@@ -52,7 +52,7 @@ written.
 ## LSTM's
 
 There are two LSTM's both sub class [LSTMModel](./lstms/LSTMModel.py). Note that
-in the paper the standard LSTM is called the Tweeked LSTM in the code base sorry for any 
+in the paper the standard LSTM is called the Tweeked LSTM in the code base sorry for any
 confusion.
 
 [Early Stopping LSTM](./lstms/EarlyStoppingLSTM.py) as the name suggests does not have a set number of times
@@ -77,28 +77,14 @@ more relevant this was used as it appeared to work well for this task.
 
 Require:
 1. Python 3.4.3 or above.
+2. graphviz
 
-If you would like to visualise the LSTM's then GraphViz is required for Debian based
-systems this can be installed using:
-
-apt-get install graphviz
-
-### Note on [Unitok-3.0.3](./unitok-3.0.3)
-
-I have included unitok-3.0.3 within this project as this project requires a Python 3
-version and the one currently [available](http://corpus.tools/wiki/Unitok) is
-Python 2 only therefore this version is Python 3 only for English.
-
-To install go to [Unitok-3.0.3](./unitok-3.0.3) folder and run:
-
-python3 setup.py install
-
-### All of the other pips
-
-All the other pips can be installed using the following command:
+And the installation of pip's:
 
 pip3 install -r requirements.txt
 
+Also look at the [config file](./config.yml) to see where to put the data.
+
 
 ## [Final output](./final_output)
 

results/best_aspect_clf_results.tsv

@@ -1,2 +1,2 @@
-Mean	SD	union__ngrams__posextract__expand	union__ngrams__posextract__replacement	union__ngrams__negextract__replacement	union__ngrams__negextract__expand	union__ngrams__count_grams__binary	union__ngrams__posextract__expand_top_n	clf__C	union__ngrams__negextract__expand_top_n	union__ngrams__text_extract__feature	union__ngrams__tokeniser__tokeniser_func	union__ngrams__compextract__expand	union__ngrams__negextract__words_replace	union__ngrams__compextract__words_replace	clf__epsilon	union__target_extract__count_grams__binary	union__ngrams__posextract__words_replace	union__ngrams__compextract__replacement	union__ngrams__tokeniser__ngram_range	union__target_extract__aspect__feature 
-0.617100347856	0.0460432874772	Word2Vec(vocab=38074, size=300, alpha=0.025)	posword	negword	Word2Vec(vocab=38074, size=300, alpha=0.025)	True	10	0.1	10	text	unitok_tokens	None	Poor word	train companies	0.01	True	Excellent word	companyname	(1, 2)	aspects
+Mean	SD	union__ngrams__negextract__expand_top_n	union__target_extract__count_grams__binary	union__ngrams__compextract__replacement	union__ngrams__tokeniser__tokeniser_func	union__ngrams__negextract__replacement	clf__epsilon	union__ngrams__compextract__words_replace	union__ngrams__text_extract__feature	union__ngrams__posextract__replacement	union__ngrams__count_grams__binary	union__ngrams__negextract__words_replace	union__ngrams__posextract__expand_top_n	union__ngrams__posextract__expand	union__ngrams__posextract__words_replace	union__ngrams__negextract__expand	union__ngrams__tokeniser__ngram_range	clf__C	union__target_extract__aspect__feature	union__ngrams__compextract__expand 
+0.617105595922	0.0460405648102	10	True	companyname	unitok_tokens	negword	0.01	train companies	text	posword	True	Poor word	10	Word2Vec(vocab=38074, size=300, alpha=0.025)	Excellent word	Word2Vec(vocab=38074, size=300, alpha=0.025)	(1, 2)	0.1	aspects	None

results/best_clf_results.tsv

@@ -1,2 +1,2 @@
-Mean	SD	posextract__words_replace	negextract__words_replace	posextract__replacement	posextract__expand	negextract__expand_top_n	negextract__expand	negextract__replacement	count_grams__binary	tokeniser__tokeniser_func	compextract__words_replace	compextract__expand	clf__C	clf__epsilon	compextract__replacement	posextract__expand_top_n	tokeniser__ngram_range 
-0.614559367748	0.0468039247619	Excellent word	Poor word	posword	Word2Vec(vocab=38074, size=300, alpha=0.025)	10	Word2Vec(vocab=38074, size=300, alpha=0.025)	negword	True	unitok_tokens	train companies	None	0.1	0.01	companyname	10	(1, 2)
+Mean	SD	negextract__words_replace	compextract__replacement	clf__epsilon	posextract__replacement	compextract__words_replace	posextract__expand_top_n	negextract__expand_top_n	posextract__expand	compextract__expand	negextract__expand	tokeniser__ngram_range	count_grams__binary	negextract__replacement	posextract__words_replace	tokeniser__tokeniser_func	clf__C 
+0.614564261783	0.0468012629474	Poor word	companyname	0.01	posword	train companies	10	10	Word2Vec(vocab=38074, size=300, alpha=0.025)	None	Word2Vec(vocab=38074, size=300, alpha=0.025)	(1, 2)	True	negword	Excellent word	unitok_tokens	0.1