Extract information from texts

Recall : The transformation Extract information from texts is possible for the text variables.

Choose the type of variable you wish to create : lexical, lemmatised, analysis via a thematic dictionary.

Lexical variable

You can generate different types of variables having the words chosen as possible modalities (to be separated by « ; »). In our example, we choose to consider the three words more frequently cited  (health, life and well-being) to the question « Picture_comments » which asks respondents to cite the words coming to their minds when seeing the wall of images.

The lexical measures will create variables allowing to analyze the content of your text variable. The length corresponds to the number of words of the response (if the answer is « life is beautiful », the length will be 4, if we take into account the words tools), the richness is the number of different words in one response, the banality is an average of frequency of the words of a response (The more frequent the words are the more the banality is), finally the intensity gives the « rate of presence » of the words chosen (in our case "health", "life" and "well-being") in a response. For example, if the answer is « Life », the intensity of the word « life » will be 50%, as the response contains two words one of which is « Life » (if we take into account the tool words).

Lemmatised variable

This assistant allows to create :

• text variables which contents will be the texts of the selected variables in their lemmatised form (infinitive, masculine, singular) : example, « he drinks » becomes « he drink ».
• several text variables for each grammatical category of the selected variable (verbs, nouns, adjectives) : example, « I drink water », one variable Verb will be created containing the value « drink » and one variable Noun containing the value « water ».

Thematic analysis

From a dictionary of words stored by themes, to load by clicking Choose a thematic, this assistant will create an ordered variable of the more cited themes as well as the closed variables corresponding to each of the themes.

Before finishing, select the location of insertion of the new variables.