Add an API to feed metadata in and produce valid codemeta/schema.org

[rebeccaringuette]

Requesting an API be added to the auto-codemeta generator for generating codemeta jsons in large quantities. The input would need to be some basic json where a program can send the information in some basic json format (not codemeta) and the return would be validated codemeta json. I am finding we need this in several points of operations but there is no available way to do this.

• use case: Auto-generating valid codemeta json from a database of software metadata. The database has its own terminology and structure but needs to generate codemeta. All extractions from available information have already been performed.
• use case: auto-generating schema.org aligned codemeta for auto-embedding that metadata in a software landing page to increase discoverability by generic internet searches (potentially also AI searches). This must be automated to ensure the embedded metadata is always accurate compared to the metadata stored in the database. The input metadata comes from an existing metadata database.

These use cases call for an API with a simple JSON input structure and two types of output structures. The input structure should not be the codemeta structure (otherwise there is not much point to calling the API to generate codemeta json) and should basically allow input to any field in the generator. The output structure should have two options, either valid codemeta or valid schema.org-aligned codemeta. Having this as an API centralizes the maintenance and governance of what is valid codemeta/schema.org for software metadata in the correct group (CodeMeta) and provides a standardized machine-actionable way to create this metadata.

A potentially second API could be called in the background by the first API to validate the desired codemeta/schema.org flavor of metadata produced. This separate API would also be useful for external services to call to check that existing codemeta/schema.org metadata is valid compared to any updates since the metadata was generated.

[dgarijo]

At the moment the only API available receives a repo and generates a CodeMeta.json file.

We will be expanding this so an alternate endpoint will generate CodeMeta compliant with schema.org.

This other API request is a little confusing. Is it specified anywhere? Are there examples we can see?

The validation APIs, we can see. At the moment they are done in the website, not exposed.

[rebeccaringuette]

The idea is to feed in metadata into an API and the output would be correctly formatted codemeta compliant with schema.org. Something like an API version of the generator interface.
Alternatively, this could be a code like the soso package (https://github.com/clnsmth/soso) where various source metadata structures could supply a maping from their terms into codemeta and the software package produces the desired codemeta flavor. However, such an approach is likely time-consuming compared to an API.

[dgarijo]

Sorry for the delay in my answer.
I think this is a like a snake biting its own tail: for the API to exist, you need to send data following a JSON format. And having everything in a JSON format is what we aim to achieve, no? (that is the codemeta file).

The code that integrates several metadata formats in CodeMeta format is somef (or at least that's the idea). The google compliant version with -gc is ready, but needs to be merged: KnowledgeCaptureAndDiscovery/somef#888
Maybe that goes in the right direction?

[rebeccaringuette]

No, we need an API to give information for each field to, which the service would then format correctly into CodeMeta / schema.org-aligned form and give it back. The correct structure of the information is not always straightforward (e.g., funder and funding, role and contributors). The input API could definitely use the codemeta terms, but the correct header, structure of the objects, and so would be done by the service.