Dragon Text Mining PWMs Generator  
How to use the system?

Dragon Text Mining PWMs Generator (DTMPG) can generate Position Weight Matrices (PWMs) from free text sentences submitted by users. You can follow these steps in order to use the system:

Step 1: Provide input data to the system.
  • You can upload a ".txt" file that contains the text. Then provide keywords that will be used to generate the PWMs (maximum three words). Then click "Submit" button.
  • Alternatively, you can click "Sample Input Data" button for sample data.




Home Page
Then you will be redirected to another page. You will find the sentences that include all the keywords, and the keywords will be highlighted.

Input Data

Step 2: Click "Generate PWMs" button. This action allows the system to process your input data and generate the corresponding PWMs according to the provided keywords.

The Generated PWM

Step 3: Click "Normalize PWMs" button. This action allows the system normalize the PWMs according to the selected normalization mechanism.
  • The first normalization mechanism is computing Pi,j, which is the probability that word i appears in column j.
  • The second normalization mechanism is computing log(Pi,j), which computing the natural logarithm of the probability.
  • The third normalization mechanism is computing Pi,jlog(Pi,j), which is multiplying the probability with the natural logarithm of the probability.





Selecting a normalization mechanism is required if you want the system to perform this action.

The Normalized PWM

Step 4: Click "Compute Scores" button. This action allows the system to use the normalized PWMs to compute the matching scores of sentences.
  • Select "Using input file" option if you want the system to compute scores for the sentences in the input file.
  • Otherwise, select "Another file" option if you want the system to compute scores for another set of sentences. In this case you should upload another ".txt" file.



The Computed Scores

DTMPG provides functionality to download the PWMs, the normalized PWMs, and the matching scores of sentences.

For any comments or technical questions, please contact: arwa.binres@kaust.edu.sa

© 2013 King Abdullah University of Science and Technology. All rights reserved