starting SIDE
Execute “run.bat” file. (It's at the SIDE/SIDE_src/SIDE/)
Double-click on SIDE.jar file will also run the program,
but it will eventually cause memory deficiency problem while training your machine learning algorithm.
The following is the first screen you will see whn you run SIDE.
(Or slightly different following your version)
import unstructured documents
- Right click on Imported Document folder at the left panel. (Unstructured Document folder following the version)
- Click on 'Import Unstructured Document'
- Choose source directory as the source directory of computer's file system.
- Choose destination directory as the destination directory of SIDE file system.
- Click 'Import Now!' and close the dialog.
- The files should appear on the SIDE file structure on the left panel of the program.
filter
Filter is the basic component in creating conditions for your summary.
Creating filter
Any of the following can create filter.
- Click on Menubar, File->New->Filter.
- Click on Filter Icon on Toolbar
- Right click on ‘filters’ folder or any subfolder.
Name your filter
Click on the folder you want your filter to be in and type in filter name. Click OK.
Filter will show up in the File Structure Panel on the left side of SIDE.
Select and Annotate training documents
Select training documents
- Click on ‘+’ icon.
- Select the document you want to train, and click OK.
segment a document
- Select document you want to segment from the document list in the training document collection.
- Select the segmeter from Segmenter panel.
annotate a document
define annotations
- Click on ‘+’ icon.
- Input name of Annotation
- Click on Color rectangle.
- Choose your color.
- Click update.
annotate a document
- Select document you want to annotate from the document list in the training document collection.
- Select segments you want to annotate. (Shift, Ctrl key works)
- Right click on selected segment.
- Select annotation you want.
Select Feature Classes
Choose Feature classes you want to extract in order to create summary.
- Click on ‘Feature classes’ tab.
- Select appropriate Feature classes.
Select Evaluation Metrics
Choose Evaluation Metrics you want to extract in order to create summary.
- Click on ‘Evaluation Metrics’ tab.
- Select appropriate Evaluation
Train model
Train your filter model in order to apply the model on the target document
- Select MLA plugin.
- Train model
Training takes some time and there is not an indicator yet that tells you training is going well.
Training with NaiveBayes should not take more than several minutes.
Summary file (SIF file)
SIF file is the setting of various options you set for summary.
Creating SIF file
Any of the following can create SIF file.
- Click on Menubar, File->New->Summary.
- Click on Summary Icon on Toolbar
- Right click on ‘summaries’ folder or any subfolder.
create/configure text recipe
A SIF file consists of multiple test/visual Recipe. A summary consists of multiple recipes whose results are concatenated at the end of the summarization process. The availability of having multiple summary enables a Summary to have different summarizing options depending on each part of summary.
Define Expression
Expression is like WHERE clause of SQL. It defines which kind of annotation you want to have in your summary.
- Click on pulldown arrow.
- Choose ‘And’, ‘Or’, or ‘Is’.
- In case of ‘Is’, click on ‘…’ button and choose annotations you want to include in your summary.
Set options
Set options of summary such as Ranking, Limiting, and Reordering.
create/configure Visual Recipe
Visual recipe is for visual summary of analyzed file.
For now, visual summary works only csv file, with filter trained on same csv file.
Run SIF
- Click on ‘Run Summarization Task’ tab.
- Click on ‘+’ button.
- Choose target document you want to summarize
- Run by clicking on ‘Run’ button.