Course II: Data Liberation and Publishing Tools

Extracting data can be performed at different levels, by carrying out individual extraction from a single paper using the tool GGI, by placing different queries to the system and produce analytical outcomes and reports, and by using the matching service to link material citations to specimens.

Individual extraction

Once the software is installed, you are ready to use it. At this point, there are several steps to follow to extract annotations and attributes from a paper using GGI.

The tutorial below will guide you on how to extract annotations and attributes from a paper using GGI. Firstly, it shows the good practices you may take into consideration before starting an individual extraction. After that, you might be ready to start extracting information!

Follow these quick steps or see the full tutorial below:

1. Detect Document Structure
2. Add Document Metadata
3. Parse Bibliography
4. Mark Bibliographic Citations
5. Mark Taxon Names
6. Mark Taxon Keys
7. Mark Treatments
8. Treatment Structure
9. Mark Treatment Citations
10. Mark Materials Citations
11. Parse Materials Citations, and finally
12. Select Zenodo License

Queries and Data Reuse

At this stage, it is important to realize how to query the Treatment Bank and analyze or reuse your extracted data. For that you may need to access a quite diverse array of repositories and platforms, where the different biodiversity-relevant (types of) data might be deposited, including:

• Treatment Bank

• Zenodo

• GBIF

• SynoSpecies

• Ocellus

You may understand the process better by going through the tutorial below. See also below a practical exercise designed to give participants hands-on experience with the PLAZI workflow, specifically focusing on the extraction and annotation of taxonomic treatments using GoldenGATE Imagine.

Additionally, you may explore some ways to produce and analyze aggregated figures and data sets by accessing some major analysis tools such as those provided by Plazi: 

• Article stats
• Treatment stats
• Treatment Details Linking statistics -TDL stats

The Matching Service

This section describes the key functionalities of the eBioDiv Matching Service that is used for linking material citations to specimens.

To facilitate the task, a semi-automatic matching approach is used. Users are presented with lists of possible matches. Once a sufficient number of matching decisions have been recorded by the users of the eBioDiv Matching Service, it will be possible to start exploring ways of gradually replacing the current semi-automatic matching approach with a fully automatic matching algorithm for data pairs with a sufficient number of data points of score high enough to warrant automatic acceptance of a match.

Furthermore, it might become possible to automatically sort out data entries that are clearly deficient and therefore unfit as an input for matching decisions. These data entries could then be relegated back to the maintainers of the data along with a request to improve the data.

As a result, fewer matching suggestions will have to be curated by humans. It will be interesting to observe how the intellectual work carried out by the users of the eBioDiv Matching Service will evolve as a result of this shift in the balance between the tasks relegated to computer algorithms and the tasks remaining with human contributors. Below you will find a use case showing how the service might be used in a real-world scenario.

You may access the link above to the eBioDiv Matching Service, to obtain a more detailed introduction to the matching process and to get access to detailed documentation related to the use of the Matching Service.

• For new users, it is highly recommended to have a closer look at the following documentation for grasping the to the matching process:

1. Introduction to the Matching Service
2. Data Source
3. Matching Algorithm

• For advanced users the key steps of the process are described in the documentation stored in the folder below.