General tab - Properties of Themes Locator window

This tab enables you to assign a table model so that you can map the theme extraction results to a table. This tab has the following options.

Table Settings

This section has the following settings.

Table model

Select an existing table model from the list or create a new model for this locator by clicking Table Models.

The results returned here come from the entity extraction, so no database table is required.

Text column

From the selected table model, map a column that returns the entity text.

Confidence column

From the selected table model, map a column that returns the entity confidence.

Theme Type column

The value for this option is set to 1 if the theme is a single-word meta theme. The value for this option is set to 0 if the theme is a common theme. Meta themes are returned if a single word occurs multiple times in the extracted themes, indicating an overall document theme.

Sentiment column

From the selected table, map a column that returns the sentiment of the entity.

Filter Theme By

This group has the following options.

Confidence Threshold

This is the minimum confidence allowed for an entity to be included in the extraction results.

The value for this option is set to 70 by default.

Sentiment

You can further filter the results by specifying the types of sentiment that are returned. As you select each option, the confidence range is updated for the appropriate selection.

There are several values for this field.

  • All. Select this value to return all sentiments found on the document, positive, neutral, and negative. This is the default value for this option.

  • Positive. Select this value to return positive or good sentiments only.

  • Neutral. Select this value to return neutral sentiments only.

  • Negative. Select this value to return negative or bad sentiments only.

  • Custom. Select this value if you want to customize which sentiments to return. You can control the level using the slider.

    Invert result range

    If selected, the confidence range is reversed so that sentiments that fall outside the selected range are returned only.