A solution to the 1001 reports you have to google and regurgitate.
A small paraphraser project used to paraphrase paragraphs of text. A tongue twister I know.
-
Clone the project open up the folder iHateReports, in that folder you will see another folder iHateReports, folderception.
-
Place the text you want to paraphrase in paraphraseinput.txt. Run the code and you will find your paraphrased text in paraphrase.txt.
The accuracy of application currently depends largely on the accuracy and amount of data available. Without calling an established dictionary API, data is limited.
Reading of words with different tenses is not yet implemented, hence words with tenses that differ from the base words in the dictionary will not get paraphrased.
Cannot yet recognize degree of Word: e.g. like/love, give/bestow. They are technically synonyms but the degree of emphasis is different.
Unable to determine context, does not recognize word form e.g. verbs, noun, adjectives etc. Hence unable to determine sentence structure and appropriate synonym to use.
https://github.com/zacharytay1994/iHateReports/blob/master/exampleone.PNG
It's very primitive as per the limitations.
Dictionary data taken from http://www.gutenberg.org/ebooks/51155. References can be found in 51155-0.txt.
-
Implement reading of words with different tenses to reduce the amount of data needed (instead of treating the same word with a different tense as a different word).
-
More punctuation consideration.
-
Allow for phrases to be synonyms.
-
Allow for phrases to be keys, a different method of reading text to accommodate this.
-
A priority system for synonyms, most used > least used.
-
Semi-automatic paraphrasing option (with some degree of control)
- Sentence structure rules/recognition
This is just a fun little project to practice my basic c++ proficiency. Feel free to clone/fork/reproduce at your leisure, preferably fork and contribute here!;D