Develop an application/platform to Extract Handwritten Text from Image to a doc format will NLP/BERT analysis.
Budget: $30 – $250 USD
I have a pdf document that contains students' responses and a series of scores
1- The first step will be to transcribe the responses into a spreadsheet through an automated text reading process. Then the student's text will be accompanied by the teacher's assigned scores for that response.
2- I interested in analyzing text data with pre-trained models like BERT.
3- Representing the text as a bag-of-words n-gram model (probably unigrams, bigrams, and trigrams would be sufficient), and then compare the two approaches.
4- Report
-------------------------------------------------
In other word:
So I would like to develop an application for teachers that would allow them to automatically score and provide document summarization for students' open-ended written responses on assignments and tests. The procedure should be pretty straightforward:
(1) Obtain the open-ended responses and accompanying rubric scores
(2) Use bag-of-words or a pre-trained model like BERT to convert the text into features
(3) Train a supervised algorithm to predict students' scores based on those features
(4) Test the algorithm on new open-ended data that come in.
(5) Explore how the features of the text change as students' scores change. This could involve document summarization techniques, sentiment analysis, parts-of-speech analysis, and more.
(6) Ultimately, I'd like for you to develop an app or macro that a teacher can use to automatically grade and summarize open-ended responses.
(7) Complete Report/Documentation about every single step.
1- The first step will be to transcribe the responses into a spreadsheet through an automated text reading process. Then the student's text will be accompanied by the teacher's assigned scores for that response.
2- I interested in analyzing text data with pre-trained models like BERT.
3- Representing the text as a bag-of-words n-gram model (probably unigrams, bigrams, and trigrams would be sufficient), and then compare the two approaches.
4- Report
-------------------------------------------------
In other word:
So I would like to develop an application for teachers that would allow them to automatically score and provide document summarization for students' open-ended written responses on assignments and tests. The procedure should be pretty straightforward:
(1) Obtain the open-ended responses and accompanying rubric scores
(2) Use bag-of-words or a pre-trained model like BERT to convert the text into features
(3) Train a supervised algorithm to predict students' scores based on those features
(4) Test the algorithm on new open-ended data that come in.
(5) Explore how the features of the text change as students' scores change. This could involve document summarization techniques, sentiment analysis, parts-of-speech analysis, and more.
(6) Ultimately, I'd like for you to develop an app or macro that a teacher can use to automatically grade and summarize open-ended responses.
(7) Complete Report/Documentation about every single step.