Inspiration
Our inspiration was our dissatisfaction with Adversarial Context Machine Emergence (ACME) systems or other submission similarity comparison algorithms. Having to restructure original ideas in order to satisfy the requirements on previously submitted papers.
What it does
The application restructures given text in an attempt to decrease the similarity report given by ACME systems.
How I built it
The program was built using python with the libraries spaCy and nltk (natural language toolkit) with webnet. The idea is to use spaCy to identify the common nouns and verbs in our text segment and use nltk to determine the most appropriate synonym to replace the subject noun or verb.
Challenges I ran into
Finding a python library that would retrieve synonyms took some trial and error. Outdated python libraries made it difficult to progress through the project. We tried 4 different thesaurus lookup libraries that didn't give us the results that we wanted but ended up finding the natural language toolkit that gives us accurate synonym lookups. We wanted to use the
Accomplishments that I'm proud of
What I learned
What's next for Ma'ii
Using an offline thesaurus would greatly decrease the runtime of our program and would work indefinitely. The program currently relies on the natural language toolkit and webnet servers to provide our thesaurus lookup. We also want to implement the Wu and Palmer semantic Relatedness algorithms to grealty increase the accuracy of replaced synonyms and use a statistical reference point instead of randomly picking a synonym with the same word type.
Log in or sign up for Devpost to join the conversation.