BLISS

An artificial language incorporating both syntax and semantics. This code was used to generate results of our publication on constructing an artitifical language of intermediate complexity, which is suitable for language learning in neural networks. BLISS mimics natural languages by having a vocabulary of ~150 words, a context-free grammar, and some semantics, as defined by a degree of non-syntactic statistical dependence between words.

Code description

Main files:

• pybliss.py: The main file, which you need to run to generate sentences
• cnstpybliss.py: The constant file, where you can set the input and output file names and parameters

Auxiliary files that pybliss.py uses for sentence generation:

• blissutil.py: Contains the utility function for working with dictionary and for reading different file types in pybliss.py;
• blissplot.py: Plotting functions for different file types
• KLdiv.py: Calculates Kullback–Leibler divergence between two different distributions
• CountFreq.py: Calculates the frequency/probability of all words/pairs inside a corpus

Generating sentences

1. Set the parameters in the constant file, cnstpybliss.py:

• FILE_NAME: the prefix for the name of the output files
• GRAMMAR_FILE: the grammar file name
• JNTPROB_FILE: the joint-frequency file name
• UNIGRAMPROB_FILE: the word-frequency file name
• WORDROOTS_FILE: the root of words
• NUM_OF_SENTENCES: Number of sentences to be generated
• PRINT_SENT_No: Print on the terminal after these number of sentences were generated
• SELECTION_ALGORITHM: select the language model out of ALL_ALGORITHMS, which are different language models described in our paper

2. Run the main python file:

python pybliss.py

Then, you find the generated sentences stored in FILE_NAME with txt format in the current directory.

Developer

This package is developed by Sahar Pirmoradian. If you need help or have questions, don't hesitate to get in touch.

Citation

If you use this code, please cite the corresponding paper where BLISS was introduced:

S. Pirmoradian, A. Treves, “BLISS: an artificial language for learnability studies”, Journal of Cognitive Computation 3:539—553, 2011

For getting a broader picture about the project and how this artificial language could be used, please see my PhD Thesis, where I used the BLISS language as the training language of an associative memory network.

Acknowledgments

• This project was done under the supervision of Prof. Alessandro Treves, in SISSA, Trieste, Italy.