Scones: Story Tagging

From TechWiki

Introduction

The scones (subject concepts or named entities) tagger is how you identify and then tag information in document and text content in accordance with the ontology or named entities common to your domain.

This document explains how users should use the structScones conStruct module to tag concepts and named entities, which are part of the structWSF node, to different input documents; and then to index this tagging in the system so that they because available to users of the node.

The scones process is composed of four steps:

1. Submit Story
2. Review Tags
3. Adding Additional Metadata
4. Done

Logging In

This function is generally limited to those with administrator or similar privileges. You must be logged in to use this system.

Invoking scones

The user interface for scones may be accessed directly:

• http://your-domain-name.com/conStruct/scones/

Or, you may get to it via the conStruct tools listing in your specific instance:

(The look of this and its location may vary for your instance.)

1. Submitting Stories (text content)

The first step consists in submitting the story to the scones tagging system. Right now, only plain text documents can be submitted directly to the system, but it is planned to support more file formats in the future, such as PDF, DOC, etc. For now, all these other file formats have to be converted into plain text in order to get tagged by the system.

You have to specify a title for the story you are submitting, and you have to copy/paste the text of your story into the "Enter or Paste Story" section.

Once this is done, you have to click the "Submit" button.

2. Tagging the Story

The next step consists in reviewing the tags that got tagged for your article. The text, and its tags, will be displayed in a story viewer tool. This story viewer tool shows you what has been tagged in your article.

Each possible tag is listed in the right panel with a checkbox. Clicking an entry causes it to be highlighted in the story (and also committed as a tag for the story or document when saved).

Because of inferencing, you may find terms in the right panel that are not specifically included in the actual text. When you click one of these entries, its inferred parent is what is then highlighted.

If you are OK with the tags, you will process with the next step by clicking the "Accept" button. If not, then you will click the "Refuse" button.

Right now, there is no way to modify the tagged terms using the Story Viewer. This feature should be implemented later. For now, the only way to modify the tags is by using the GATE Developer user interface.

3. Adding Additional Metadata

The next step is a way for the system to ask more information about the story to the submitter of the story. This step changes from one system to another, depending what additional information is needed by the organization that collects these stories.

This step only consist in filling the blanks with what you know related to the requested information.

4. Done, and More

This step is a confirmation that everything went well with your story submission, and then asks you if you want to see the document you just tagged, or if you want to tag a new one.

All tagged documents are indexed in a dataset specified by the system administrator.

Annotate the document allows you to begin the process anew by returning to Step #1.

Changing the Tags/Deleting the Story

After submission, you may later want to amend or delete your submission.

Though there are multiple access points for which to do make changes, the easiest is to use the conStruct search function. It is often accessible via the same conStruct tools listings shown at the beginning of this article.

Once accessed, you will get the search box. Assuming you have the administrator rights to the dataset for your record (a safe assumption if you tagged the document in the first place), enter a unique or descriptive term for the record (document) you want to alter:

You will get a listing of matching results (in this case, only one).

With admin rights, you will see some icons to the right of the entry. One allows you to view or modify the record; the other (which is highlighted) allows you to delete the record. For this case, we will choose delete.

That action then brings up the delete confirmation:

Note the record identifier shown; this is often useful for other editing or modification needs. If you confirm the delete, the record will then be permanently deleted.

(Note we could have chosen the alternate path, which would have allowed you to view and modify the record.)