Skip to content

Latest commit

 

History

History
77 lines (56 loc) · 2.56 KB

README.md

File metadata and controls

77 lines (56 loc) · 2.56 KB
Raw

depression_classification

Lexicon-based method to detect signs of depression in text

DESCRIPTION

This software contains an unsupervised method to detect signs of depression. It takes a text and returns a value from 0 to 1 by taking into account a domain-specific lexicon. The classification method is a baseline strategy based on identifying entries of the lexicon in the input text. Two versions of the lexicon are used: both just word forms and lemma / PoS tags pairs. The specific lexicon provided by this repository is that described in Choudhury et al. (2013).

The repository is also provided with other lexicons expanded with the help of distribution models and WordNet from the two existing ones: Choudhury lexicon (Choudhury et al. 2013) and Pedesis lexicon (Neuman&Cohen 2015).

All lexical resources, including expanded lexicons, are in folder ./lexicons. More information can be found in the following paper:

Losada, David and Pablo Gamallo (...) "Evaluating and Improving a Lexical Resource for Detecting Signs of Depression in Text"

REQUIREMENTS

  • Linguakit

Please, donwload Linguakit:

https://github.com/citiususc/Linguakit/archive/master.zip

and copy the Linguakit folder to ./depression_classification at the first level.

HOW TO INSTALL

You only need to download the repository.

ZIP Download

Download depression_classification-master.zip and then:

unzip depression_classification-master.zip

Using Git

git clone https://github.com/gamallo/depression_classification.git

HOW TO USE

Run ./classif --help to see the modules:

echo "Syntax: cat input_file |./classif  <module> [PoS_tags]
      
      module= tok or tag | you can just use a tokenizer (tok) or a PoS tagger (tag)
      PoS_tags= N, V, J | if you use module tag, it is possible to classify using a specific tag: noun (N), verb (V), adj (J), or a combination of them: e.g. NJ, VNJ, JV, ... (by default: NVJ).

      example of use: 
      echo "I hate my life" |./classif tag VJ

EXAMPLES

Tokenized text:

echo "I hate my life" |./classif tok

Tagged text without specifying any PoS tag:

echo "I hate my life" |./classif tag

Tagged text with just nouns:

echo "I hate my life" |./classif tag N

Tagged text with nouns and adjectives:

echo "I hate my life" |./classif tag NJ

Tagged text with nouns, verbs and adjectives (by default configuration):

echo "I hate my life" |./classif tag NJV

(all combinations of these 3 PoS tags are possible)