Introduction Edit

Essay Helper is a GitHub Repository written in SWI-Prolog that helps write the introduction, exposition, critique and conclusion of a humanities essay given an essay to write on. The algorithm uses KNN to detect reused sentences, for example in comments and connections. It uses a positivity checker that requires all sentences to contain only positive terms, sometimes except critique comments and connections. 

The algorithm repeats asking for reason data if the user enters data that fails, for example if it can't recognise reused phrases. 

KNN Limitations Edit

The algorithm uses the K-Nearest-Neighbour algorithm which recognises the identical phrase and the phrase with only the first word changed, but fails if replacements or deletions involving multiple words are made at the start of the phrase. It can fail when trying to recognise the phrase with only the first word changed if there is another change. 

Positivity Checker and Implications Edit

Because the essay helper requires positive terms in the exposition, the user can't fail the assignment. Because the essay is correctly structured, the user can't earn a mark lower than D (65%). Because the user must breason out and hand in the first 80 breasonings, the user can't earn a mark lower than B (75%). The user will earn 80% if he agrees and 75% if he disagrees. 

Future Features Edit

  • Mind Reading and Random writing of essays 
  • Changing from KNN machine learning to an algorithm that captures all possible paraphrasings 
  • File output of essay, automatic breasoning (outputs breasoning dictionary only for the essay) and marking. 
  • Automatic output of essay traversing the argument in a tree structure.

References Edit

Results of Various Tests of K-Nearest-Neighbour to Recognise a Paraphrased Statement

Community content is available under CC-BY-SA unless otherwise noted.