AlchemyAPI extractors
Wandora contains several AlchemyAPI extractors. Wandora's AlchemyAPI extractors take an input, usually text, and distills topic map fragments i.e. topics and associations out of the provided text using AlchemyAPI. Wandora's AlchemyAPI extractors are
- Alchemy entity extractor
- Alchemy keyword extractor
- Alchemy category extractor
- Alchemy language extractor
- Alchemy sentiment extractor
- Alchemy image keyword extractor
Available extractors reflect the structure of AlchemyAPI services. To use Wandora's AlchemyAPI extractors, you need a valid API Key for the service. You can request your personal API Key at [1]. Wandora doesn't store your API Key between use sessions. When you exit the application, Wandora forgets your API Key.
Wandora's AlchemyAPI extractors locate in File > Extract > Classification. Addition to the file menu, extractors can be used in occurrence editor also. Occurrence editor's extractors start with menu options in Refine > Alchemy. Occurrence editor's AlchemyAPI extractor use selected text as the extractor input.
AlchemyAPI extraction example
This chapter shows you how to actually use Wandora's AlchemyAPI (entity) extractor. First you need to start the extractor with menu options in File > Extract > Classification. Choose Alchemy entity extractor....
A dialog opens with three tabs. You can extract entities from (text) files and urls resolving (text) files. Third tab is used to inject actual text to the extractor. Select third tab with title Raw and paste some text to the text field. In our example we have pasted Wandora's description text to the field. Press Extract button.
If this is first time you execute AlchemyExtractor, Wandora asks your personal API Key for the AlchemyAPI. You can request your API Key here. Enter your API Key to the field and press OK button.
Now Wandora sends your text to the web service of AlchemyAPI and receives an XML feed with extracted entities. Then Wandora creates topics for the text fragment and received entities and associates entity topics with the text document topic. After extraction you should see extraction logs as shown below.
After successful extraction you can open the text document topic created for the given text fragment and see entity topics associated to it. In our example AchemyAPI gave us two named entities: Java and Wandora. If you look closer at each entity topic, you might notice an entity topic has been classified with type topics.
In general, the quality of extracted entities varies a lot. For example, in our example the API didn't recognize entities such as Topic Maps or GNU or GPL.
Additional notes
Wandora wraps the extracted text to an URL parameter and sends the URL request to the web service using POST method. Therefore AlchemyAPI extractors usually accept relatively long texts also. However, AlchemyAPI itself has a text length limit and Wandora user can't extract any length texts with AlchemyAPI extractors.
