User:Fako Berkers/emochain

From XPUB & Lens-Based wiki


Emo chain

I didn't attach a new medium to the chain code. Instead I worked with the filtering of emotions out of a romantic novel text [1]. In the near future I want to perform the emotions that are generated by the chain. I will be acting to be in love with audience members while at the same time experiencing the instructed emotions. This way the system will simulate a person being in love that goes through different emotions and phases of love. The audience will get to experience that someone is in love with them fed by a wide range of constantly changing emotions. It is my intention that this simulation will give a generated experience. In the future narratives may be generated through simulations in a similar way.

Technically not much has changed since the Markov chain assignment. I just installed a filter upon the input text that makes sure only emotions are part of the chain. I got a list of emotions from a site [2]. From this list I took out the emotion "open", because it was triggered too often since a sentence like "He looked through the open door and saw ..." will falsely match this emotion. You can test the chain by downloading the emotion list, the sample chapter I used and slightly altered Markov chain code in one package here. Below is sample output containing the dictionary holding the chain and a random read out.

# the chain saved in a dict object
{'pity': ['humiliation'], 'humiliation': ['weary'], 'angry': ['scared'], 'sarcastic': ['glum'], 
'weary': ['pleased'], 'grief': ['sorry'], 'shame': ['sadness'], 'sadness': ['sadness', 'grief'], 
'glum': ['happy'], 'serenity': ['angry'], 'sorry': ['shame', 'sorry', 'pity', 'serenity'], 
'pleased': ['sorry'], 'scared': ['sarcastic'], 'happy': ['happy', 'EOF;']}

# output from a random read out of the chain
sorry shame sadness sadness sadness grief sorry serenity angry scared sarcastic glum happy happy happy happy happy happy happy happy happy happy