Collections and the API Integration
If you need a general idea on how we function. Watch this, it only takes 5 Minutes to bring you up to speed.
This guide deals with deciding on the number of collections to use during your work with the AX NLG Cloud.
Reasons for Multiple Collections
If you need to distinguish between certain groups within your data, different collections are your tool of choice:
- Each collection can individually be filled with a subset of your data.
- Each collection can then be individually triggered to generate content for the contained data subset.
- The collection is then individually sending their content back to your web platform together with the generated content, the collection id, the collection name, the document id, the document name and the language of the content generation.
Further reasons for creating multiple collections:
- Separate content generation for testing or for your productive usage.
- Content generation in different languages.
- Content generation for different publishing platforms.
- Content generation for a specific range of your data which is treated differently than the rest (either in generation rhythm or content wise).
These cases require the usage of additional collections for your projects, in contrast to sending all data into the same collection. The next step would be for your to decide how your data should be organized within the AX environment.
5 Relevant Factors for Your Collections ...
- Product category/topic (shoes, toothbrushes, ...),
- Tonality (formal, casual, family oriented),
- Text part (Headline, long description, short description) -> important for your field mapping,
- Output channel (Amazon, ebay, wordpress, ...).
... and how to Transport Them in Your Collection Setup.
- Multiple languages are always split into at least the same ammount of collections.
- Product category/topic is best split into collections or projects, depending on whether products/topics can be texted together or not. Both splits work.
- Tonality, textpart, output channel should be split into collections.
How to Name Your Collections
The collection name must reflect the segmentation and contain the 5 relevant factors, so that you can assign the text on your page. In addition, the data should contain which product type, text part, output channel or tonality is desired for the respective data set, if the respective factor is necessary for a textual subdivision. If one of these factors is not desired, the collections do not have to be multiplied out. In addition, the information does not have to be at the data.
So, "shoes amazon long family oriented" works well as a collection name.
At last some hints for the collections when you want to fully automate your workflow.
These are the most important API Calls when dealing with collections:
- Create a new collection: https://developers.ax-semantics.com/v3/collections/create-a-new-collection
- Generate content for all datasets within a single collection: https://developers.ax-semantics.com/v3/collections/generate-content-for-a-whole-collection
- You can check wether a dataset is present inside a collection by using the update or create endpoint: https://developers.ax-semantics.com/v3/documents/update-or-create-by-document-uid
If you need further guidance, feel free to contact your customer success agent, so we can discuss your application of collections. Alternatively, the support chat is always available.