Word vector text inflector

[mbwolff]

Using gensim to build a word2Vec model based on over 1300 French texts from the nineteenth century, I am writing code that takes a pair of words (e.g. "homme" and "femme") and a text (Le Père Goriot, by Balzac) as parameters and generates an "modulated" text. Each word in the original text is replaced by a word that is "most similar" to it according to the word pair. For instance, if "roi" is a word in the original text, it would be replaced thusly:

>>> model.most_similar(positive=['femme', 'roi'], negative=['homme'], topn=1)
[(u'reine', 0.8085041046142578)]

Handling verb conjugations and adjective agreements in French is tricky but I aim to produce a mostly readable text. The code will hopefully be able to "modulate" any text in French against any pair of words.

[mbwolff]

And it's more or less done! Here's the repository with the input text, code, vector data and output. The generated novel is Madame Bovary Modulée, based on Flaubert's famous text.