SciCurve: revealing life science’s curves

Screen Shot 2014-09-25 at 11.00.18 AMAfter the recent launch of Sciencescape, here is  another startup pledging to help us cope with the enormous amount of data and literature at our disposal. SciCurve uses PubMed‘s library of 23 million references to generate visually pleasing graphs and curves that helps you grasp trends in literature.  It comes with three main functionalities

Observing trends for your field of study.

For academic researchers, knowing about the publication trends in a field is fundamental when writing a review or a grant proposal. Is this field new? Has is been prolific in the 70s and is coming back? Or is it the latest hot field and is in exponential growth? SciCurve saves you the trouble of going through PubMed results and manually copy/pasting the number of search results. Simply enter one or more keyword and it will display a timeline of the number of publications and citation counts over the past 14 years.

Her's an exemple of teh trend function, showing an increasing number of publication since 2000, which is followed with an exponentially increasing number of citations.

The number of publications and citations for the keyword “biomaterials”. The raising  publication count since 2000 is followed by an increase in the number of citations.

 Exploring research networks.

One particularly challenging task when exploring the literature, is to quickly find the key papers, on which most of the field is based. Manually going through papers and tracing back the original work is time-consuming, and the large amount of information gathered could end up be confusing. SciCurve helps you analyse the literature automatically and helps you understand its findings by generating representations of how paper are interconnected.

The network tab displays a map of the network of publications based on citation relationships. Important papers tend to form nodes from which radiates the papers citing that work. Key papers, and the articles citing them can then easily be identified.

Example of a networks graph, with Biomaterials as a keyword.

Example of a network graph, with “biomaterials” as a keyword.

Second, the map function places on a map top papers of the field represented by circles of size proportional to their relevance. SciCurve clusters similar papers together, naturally point-out relevant sub-fields. This is particularly useful to understand what a field might entail.

Example of a map view with the keyword "Biomaterials" Clusters form show that sub-categories such as collagen, surface and engineering are prominent.

Example of a map view with the keyword “biomaterials”. Clusters reveals  sub-fields such as collagen, surface and engineering.

Finding who drives the scientific endeavour?

SciCurve also automatically generates author profile pages, including a map of their most frequent co-authors and a list of their publication. Wondering who are the most prolific authors in your field? SciCurve can identify the key authors in a particular field and map them out as a function of their publication record.

A map of the most published authors in the field of biomaterials

A map of the most published authors in the field of biomaterials

The graphs and maps, but not the raw data, can be downloaded in the free version of SciCurve. An enterprise version, with more advanced functionalities is offered for a fee. SciCurve also supports Zotero and Mendeley integration which allows you to easily export  references to your favorite citation management tools.

Beyond a rather obvious usefulness for academics and R&D scientists, this sort of tool could be interesting for general practitioners and other medical specialists that do not have the time to grasp the latest trends in medical research. Publishers and research databases could also use this approach to improve their search engines by providing a more data-rich and more visual experience to their users.

Sciencescape offers a new approach to searching for scientific publications

20202At the pace research is advancing, researchers must keep up to date with an impressive flow of information. Millions of papers are published every year and millions more are already available. So how can we stay up to date with the latest advances in our fields? And when exploring new areas of research, how can we identify the important papers that will help grasp the field quickly? Many great online tools such as  PubChase, Scizzle and scientific search engines such as Google Scholar, Pubmed and Web of Knowledge will all help you find and collect scientific papers. And you can now count on Sciencescape as well.

Sciencescape is a new way to discover and gather scientific publications. It uses the power of full text mining and network theory to help you find papers highly relevant to your search and interests. Sciencescape was founded over 3 years ago, however their public beta version was launched just this July. So lets have a look at what this new tool offers.

Sciencescape ranks paper a bit differently

Sciencescape ranks papers not by impact factor of the journal it is published in or citation counts, but based on their Eigenfactor. This elaborate method derived from network theory, gives a more detailed representation of the impact of scientific papers by taking into account not only the number of citations, but also the impact of the paper the citation is coming from. This technique is similar to those used by Google scholar and Google search: the more links point to a particular page, and the more important are the sites linking to that page, the better ranked the page will be. Enhanced by a few customizations special to Sciencescape, this Eigenfactor-based ranking brings up lists of highly relevant results.

Sciencescape

Publications are sorted by Eigenfactor, an alternative to citation numbers.

Browse networks of interconnected papers.

Beyond paper ranking, Scienscape makes the discovery of new papers easier and more efficient. With the approval of dozens of publishers to mine the full text of their research papers, Sciencescape extracts relevant information that helps link papers together by fields. As an example, for one of my publications, describing the use of mucins glycoproteins to pattern mammalian cells on surfaces, Sciencescape identified a dozen different fields (see screenshot below). Some of these are more relevant than others, but overall, Sciencescape is doing a fine job at identifying the scientific fields behind each paper it analyses.

Screen Shot 2014-07-18 at 9.30.42 AM

Scientific fields identified by Sciencescape from the full text mining of a paper.

You can then subscribe to the fields that interest you to receive updates about the latest papers in that particular field. Clicking on the field name leads you to the field chart page. The charts represent the publishing landscape for a certain field over time. Spikes in number of citations in open access journals are easily visualized. This makes it easy to spot the papers that are likely to be important in a particular field.

Screen Shot 2014-07-18 at 9.45.00 AM

Field charts help you get a glimpse at the most impactful papers in  a particular field over time.

A paper by John Smith? What John Smith?

Properly identifying authors can be a tricky task, mostly due to the relatively high occurrence of homonymes. The ORCID project is already well underway to solve this issue and by providing each researcher with a unique ID number to track their career and publishing records. Sciencescape takes an alternative approach by compiling the large set of information they have at their disposal to properly identify authors and their list of publications. This method does not require any effort on the author’s side, so it becomes easy to explore the publication records of any of your colleagues

The author description page automatically generated by Sciencescape

The author description page automatically generated by Sciencescape.

Don’t be selfish, share your findings within Sciencescape

Sciencescape is even more powerful when used within a community (think department or lab). Papers and collections are easily sharable and can be “broadcasted” to other sciencescape users via quick tweeter-like messages. By signing up with your institute’s email, Sciencescape can identify your Institute and connect you with your colleagues. These reference collections will soon be transposable in several mainstream reference managers such as Mendeley papers and Endnote.

How much?

The service is free for citizen scientists and academics and come at a fee for researchers working in the industry. This tool is still in beta testing phase, so expect changes in the interface and the addition of many new functionalities. But it is already a well build platform with a rather innovative approach to literature search, and I recommend checking it out!

Search for facts buried in articles with EvidenceFinder

Europe PubMed Central has this interesting “lab section“, where they experiment new tools to improve how one can search their database.

EvidenceFinder is a search engine that digs deep into the full text of articles to find facts related to your search queries. EvidenceFinder then converts these facts into a list of questions. And for each question, corresponds articles containing specific answers.

For example, if I enter the keyword “mucin” (a glycoprotein, major component of our mucus), the tool will come up with a series of questions that I might have been thinking of:

  • What is bound to intestinal mucin?
  • What is bound by salivary mucin?
  • What degrades intestinal mucin?
  • What produces mucin?
  • What is inhibited by submaxillary mucin?
  • What dilutes in gastric mucin?
  • ….

Let’s say I was looking for information about what can bind to salivary mucin, the tool then displays articles that contain answers to the question. For example, here we learn that salivary mucin bins to Choleragen and “WGA” for example. The tool makes accessing very specific facts easy.

I found that the search had a very natural feel, probably because very often my search queries actually originate from questions. Usualy, I first mentally transform a question into keywords, then it is up to me to do the data mining. Here, the data mining is done for us, and the tool provides a list of questions that have been addressed in the literature. The service is still in the experimental phase, and is a bit to slow to be used in my every day search, but I will definitely be keeping an eye on this.

EvidenceFinder is developed by The National Centre for Text Mining (NaCTeM), a UK-based text mining centre that provides text mining services to the UK academic community. The tool is hosted by Europe PMC (formerly named UKPMC),  an initiative supported by 19 funders of biomedical research, including charities and government organisations in the UK, Austria, and Italy, led by the Wellcome Trust.

I will very soon create a new section in the “Online tools for researchers” page gathering the most innovative and useful research-oriented search engines.