The award-winning AI solution to understand contextless data in tables
In a nutshell
Thousands of contextless data are existing in many tables. Often, these tabular data are hardly intelligible by neophytes because some business knowledge or their producers are required to understand and use them.
- Table Annotation allows you to enrich tabular data with semantic annotations and make them more meaningful and usable for data science, data governance or knowledge graphs use cases
- As of today, its annotation capability relies on encyclopedic knowledge from open community sources such as Wikipedia, Wikidata or DBpedia
- Table Annotation is the product of a collaborative research project (aka DAGOBAH) in Semantic Web and AI from Orange in association with the academic partner EURECOM
The service is available worldwide
This API has two separate resources: /preprocessing and /annotation. This version allows you to annotate tabular data with the scope of encyclopedic knowledge.
Read carefully the online documentation. To get specific support on this API, you can contact our support team
Testing the API is free of charge with a limitation of 100 requests with a maximum peak load of 5 requests/second.
Reach us out through our support team if your are an academic institution or if your need exceeds this volume of requests.
Why you will love this API ?
It unlocks the value of your data
Use /preprocessing to have your data table correctly structured and cleaned without the nightmare of manual operations (unwilling special characters, empty and/or duplicated lines and merged cells are being handled automatically).
Magic isn't it?
It makes your tabular data usable
Once the data table is cleaned, use /annotation to have semantic information added to your contextless data so that they become context-augmented.
By doing so, they become much more search-friendly and suitable for applications based on business datasets.
It will make knowledge actionable
Context-augmented data can usefully enrich knowledge graphs*. Knowledge graphs can be based on general academic knowledge or specific to company business knowledge.
Search engine or virtual assistants would use such knowledge graphs to enrich their responses capabilities.
(*) A knowledge graph is a graph structure aggregating large data corpora in the form of nodes linked by relations that describe a character, a topic, a domain, an organization, etc. It is built by using open community contribution (e.g. Wikipedia) or company business experts inputs for their own use.
A robust and advanced semantic annotation solution
The Table Annotation solution from the team DAGOBAH has been ranked respectively #1 and #3 in the last editions of the Semantic Web SemTab challenge, amongst international competitors including renowned universities and top tier tech companies.
During the challenge, 200K+ tables were annotated with various understanding challenges and benchmarked.
It continuously improves itself by fine-tuning various dimensions such as lookup and new embedding techniques.
About the technical aspects of the solution
Check this video to understand how it works
Enjoy this 8 minutes video from the team at the SemTab2021 to see how advanced is the Table Annotation solution...
... and this short demo of DAGOBAH UI, a Web interface to manipulate, annotate and enrich tabular data.
Overview of the workflow
The workflow shall be composed, on the one hand, of a cleaning step with /preprocessing to make your table ready for annotation, on the other hand, of an annotating phase of the cleaned table with /annotation. The annotation process uses multiple annotation techniques and the help of a knowledge graph to generate additional meta data for your tabular data. For more explanation, check the video below.
Get started now
Credit image: Flaticon.com