Folksonomy 4 Sciences

Press release.

I want to become a curator!

To become a curator your email address is required. We will map your decissions to your email address. So curations you have already done are visible for you in later sessions.

Your email address:

Privacy notes

Any information submitted by users to GoPubMed is stored for internal use only. Neither of your name, e-mail address or your comments will be shared with third parties.

Scientific collaborations – Authors in biomedical literature

Biomedical research happens in networks of researchers. GoPubMed allows these networks to be searched for possible experts and collaboration partners, a feature which leads to tremendous time saving when searching for appropriate experts. This feature is especially important in a specialized scientific world where it is becoming more and more vital to set up temporary teams of highly specialized experts.

There is a lot of information on authors already present in the biomedical literature. GoPubMed presents this information in form of author profiles on top of the search result pages providing an overview on a specific author. The author's research topics and collaborators are available in the what and who section in the category tree on the left.

If the profiles need a correction

In cases where you are not satisfied with the profile information we extracted, feel free to correct it. To assure the accuracy within the edited data you need to be registered as a curator.

Editing a profile is simple:

  1. You can edit the name , email and affiliation information .
  2. Add papers or remove papers to/from your profile by clicking the add or remove icons top left of each abstract. If a paper should be missing you can run any additional searches to collect all papers for your profile.
  3. The curator mode stays open until you save your changes or cancel editing.

Become a curator

Ontology-based literature search is reliant on sophisticated text-mining. While there are ingenious techniques to reach close to human quality those techniques depend on good training data.

For an ontology-based search engine it is most important to distinguish the meaning of ontology terms in free text. This is in some cases a difficult task which needs training data for our algorithms.

What is a curator?

As a curator you can curate our search engine. That means you can edit information that will help us to further improve the search results to your questions. You are allowed to correct author profiles and provide hints for our text-mining engine. In the latter case you will mark highly relevant terms as well as terms which have a differing meaning in the abstract of articles. You will do this by simply clicking on inforamtive icons next to a list identified terms.

What do we do with the curations?

The curation data is anonymously used to evaluate and train our algorithms and by this improve GoPubMed continuously. If you see a false annotation of your article it will take very few curations to eliminate this error from all documents. By this you help improving the expierience for all users of GoPubMed and support the idea of ontology-based search.

The anonymous curation data will be available for developers of other text-mining systems as well.

What are the most valuable curations?

It’s true – more is better. However we gain more from the non-obvious curations. So if you mark rather specialized terms as highly relevant it’s much more informative than a general term.

On the opposite if you mark a more general term carefully as not correct it tells us much more than marking a specialized term as incorrect.

Rule of thumb:

It is not necessary to curate all terms of an article.

If you want to discuss/explain your decission click the discussion iconicon to submit a message.

example view of an article curation

Click the curator icon on the left (see yellow arrow) to enter the curator mode for this article. You will see the identified terms on the right side. This picture shows six curated terms for this article. In this example the abbreviation ESR was mis-interpreted as a protein.