USE CASE BiotXplorer
To search for biotic interactions in the literature:
- to identify species possibly interacting with a given species
- to identify the type of biotic interaction between two species
- to retrieve evidence supporting a biotic interaction between two species
The service preprocesses the literature to build triplets: two species and one interaction which could be optional. A few rules have been defined to refine the system: for instance, the two species must not have a parent/child relationship. All the triplets are stored in a MongoDB database which is accessible through an API and a GUI.
Need:
The service addresses a critical challenge within the biodiversity research community— the overwhelming volume of data that often hinders valuable insights. In this context, the service primary goal is to aid in the detection of biotic interactions. While its precision is dependent on many factors, the vast amount of data it processes can reveal new information and patterns. The inclusion of evidence for each triplet serves as a mechanism for domain experts to validate predictions.
Added Value:
It automatically processes large volumes of data: abstracts, full-texts, supplementary material and PLAZI treatments.
Competitive Advantage:
There are no other automatic biotic interactions builder for the moment.
A qualitative upgrade on the current use of biodiversity data:
The ability to connect databases and in particular curated data items is the main qualitative upgrade of the service. Thanks to the biotic exploration tool, it is possible given a certain interaction to search the literature as well as curated taxonomic treatment as generated by Plazi. The “literature” is here to be interpreted in a very broad sense as the system allows to connect abstracts (e.g., Zootaxa), full-text articles (e.g. Zookeys) as well as 20 millions supplementary data files, including XLS, CSV and - last but not least - OCR-ized images. e.g., phylogenetic trees, biotic interaction maps.
Exemplary Use of the Service:
This video offers some exemplary uses of the service.
Competencies and Skills that are needed to use the Service:
No specific competencies are needed. The service is available through a user-friendly interface or can be accessed from APIs. The service can be used through its user interface. A few query examples are provided. By clicking on any of these examples, it will pre-fill the forms and the user can then search and see results.
Challenges for the Users:
The service is processing a very large amount of data and might be slow to answer some queries. Moreover, in its current development stage, the triplets build englobes a number of false positives, it thus requires manual scanning of the evidence. We are currently working on an evaluation to assess the quality of the triplets and improve the model.
Users Role in the Service Development
We might add user validation to exclude irrelevant biotic interaction. This would enable us to create a training set for further development.