The 17th International Conference on Digital Libraries (TPDL2013) is almost over. There were many interesting presentations, great weather, and awesome food :-). I took some pictures, that you also find on Facebook, G+, or as a single file download on Read more…
Yesterday, we published a pre-print on the shortcomings of current research-paper recommender system evaluations. One of the findings was that results of offline and online experiments sometimes contradict each other. We did a more detailed analysis on this issue and wrote a new paper about it. More specifically, we conducted a comprehensive evaluation of a set of recommendation algorithms using (a) an offline evaluation and (b) an online evaluation. Results of the two evaluation methods were compared to determine whether and when results of the two methods contradicted each other. Subsequently, we discuss differences and validity of evaluation methods focusing on research paper recommender systems. The goal was to identify which of the evaluation methods were most authoritative, or, if some methods are unsuitable in general. By ‘authoritative’, we mean which evaluation method one should trust when results of different methods contradict each other.
Bibliographic data: Beel, J., Langer, S., Genzmehr, M., Gipp, B. and Nürnberger, A. 2013. A Comparative Analysis of Offline and Online Evaluations and Discussion of Research Paper Recommender System Evaluation. Proceedings of the Workshop on Reproducibility and Replication in Recommender Systems Evaluation (RepSys) at the ACM Recommender System Conference (RecSys) (2013), 7–14.
Our current results cast doubt on the meaningfulness of offline evaluations. We showed that offline evaluations could often not predict results of online experiments (measured by click-through rate – CTR) and we identified two possible reasons.
The first reason for the lacking predictive power of offline evaluations is the ignorance of human factors. These factors may strongly influence whether users are satisfied with recommendations, regardless of the recommendation’s relevance. We argue that it probably will never be possible to determine when and how influential human factors are in practice. Thus, it is impossible to determine when offline evaluations have predictive power and when they do not. Assuming that the only purpose of offline evaluations is to predict results in real-world settings, the plausible consequence is to abandon offline evaluations entirely.
As you might know, Docear has a recommender system for research papers, and we are putting a lot of effort in the improvement of the recommender system. Actually, the development of the recommender system is part of my PhD research. When I began my work on the recommender system, some years ago, I became quite frustrated because there were so many different approaches for recommending research papers, but I had no clue which one would be most promising for Docear. I read many many papers (far more than 100), and although there were many interesting ideas presented in the papers, the evaluations… well, most of them were poor. Consequently, I did just not know which approaches to use in Docear.
Meanwhile, we reviewed all these papers more carefully and analyzed how exactly authors conducted their evaluations. More precisely, we analyzed the papers for the following questions.
Today we released the 4th release candidate version of Docear. With all the changes introduced into Docear with RC3, there were still some bugs to fix which could cause some distortion in the workspace tree or even prevented users with Read more…
Today we released RC3 (Release Candidate) of Docear 1.0 (not yet on the official download page but here in the Blog only). It has one major change compared to previous Docear versions, namely we got rid of the “Incoming” mind map. In the past, most users never really got used to the idea why there was an ‘Incoming’ mind map and a ‘Literature & Annotations’ mind map. Now, there is only the ‘Literature & Annotations’ mind map but it has a special “incoming” node in which new PDFs are added. We hope that this concept is easier to understand. It also means that when you move a PDF from the incoming node to any other node in the mind map, and you create new annotations in the PDF, the new annotations are directly added to the PDF node in the mind map and there won’t be any new node in the incoming node. However, if you prefer the old concept, don’t worry. You can keep your old incoming mind map and use Docear as you were used to be.
Update 2013-10-14: For a more serious analysis read What makes a bad reference manager?
Update 2013-11-11: For some statistical data read On the popularity of reference managers, and their rise and fall
Update 2014-01-15: For a detailed review, read Comprehensive Comparison of Reference Managers: Mendeley vs. Zotero vs. Docear
<irony>Have you ever wondered what the best reference management software is? Well, today I found the answer on RefWorks’ web site: The best reference manager is RefWorks! Look at the picture below. It might be a little bit confusing but we did the math: Refworks is best and beats EndNote, EndNote Web, Reference Manager, Zotero, and Mendeley in virtually all categories.
Source: RefWorks
A few months ago we released Docear4Word. Docear4Word is an add-on for Microsoft Word that allows you to insert and format citations and bibliographies very easily in MS Word. Many of our users love Docear4Word. However, not all of our users are using Microsoft Word but many are using OpenOffice or LibreOffice. One of them is Stephen from Uberstudent which is a Linux distribution for learners. Stephen, as many others, urged us to develop an add-on, comparable to Docear4Word, for Libre of OpenOffice. Unfortunately, we don’t have the expertise to do this.
Therefore, we would like you to help us. Do you have experience in developing add-ons for LibreOffice and/or OpenOffice? Then, please contact us. We have prepared a description of what Docear4Libre/OpenOffice should be able to do. Read it carefully, and tell us how long you would need to implement it. And don’t forget to tell us how much money you would want for it. Exactly, we are not expecting you to do it for free. We would be willing to pay something for it. Once we found an appropriate developer we will ask our users to donate for Docear4Libre/OpenOffice and give a good amount ourselves. Also Stephen will ask the users from Uberstudent to donate.
There is a new version of Docear available for download. It’s basically the (experimental) RC1 version done right. RC2 fixes a lot of bugs that were caused by the new workspace model with multiple projects, it features a refined and polished version of the Ribbon, fixes a lot of bugs in general and supports the standard PDF viewers of MacOSX (Preview and Skim) and probably a lot of other viewers as well!
If you are still using Beta9 of Docear, a lot of things will change and improve with this new version of Docear. However converting your old maps to this new format is a one-way process (you can’t use these files with Beta9 of Docear anymore) and the process itself might take some time, depending on the size of your mind maps. Please backup your files before upgrading to Docear RC2.
Some of Docear’s new icons in the ribbon bar
Since March, Docear offers a simple web-based mind map viewer, developed with some of our volunteering students, and supported by the Freeplane team. On next Friday, July 12th, at 10:30am the students will present their final work at the HTW in Berlin. You are sincerely invited to join the presentation and be first to see Docear’s new collaboration and synchronization feature. The work is not yet ready to be released to the public but we hope to completely finish the work in the next few months. However, even the preview is really amazing! Compared to the current online viewer the new “Docear Web” offers lots of features. First of all, you can edit your mind maps online and not only on your own but together with your colleagues. The collaboration works both with your local desktop Docear and with your web-based Docear. That means, you can just start Docear Desktop as you are used to and colleagues of you may work on the same mind maps you are editing either on the Web or with Docear Desktop as well. Collaboration is in real-time, similarly to Google Docs. In addition, there is a Dropbox-like utility that synchronizes all your data between different devices (and the Web). As said, not everything is already fully functional but the preview version has at least all the basic features and gives you a very good idea what to expect for the final version.
Yesterday we welcomed Dr. Georgia M. Kapitsaki here in our main office in Magdeburg, Germany. Georgia is from the University of Cyprus and will stay one month with us. Her main interest lies in our research-paper recommender system and she will support us Read more…
The last version of Docear was released three month ago and you might wonder what we were doing. Well, I can tell you we were really busy. Besides working on some research papers for conferences in Indianapolis and on Malta (read here and here), we finally implemented two major milestones for Docear. These two milestones actually were the last ones we had on our road-map for releasing the final 1.0 version of Docear. And here it is, Docear 1.0 (RC 1) with:
We have completely redesigned the dialog that appears when Docear is first started. We believe it to be much more user friendly and intuitive. We also listened to those users who criticized that our terms of service had to be accepted even when no online services were activated. Now you have the choice. You can either use Docear as a registered user and enjoy the full potential including PDF metadata retrieval, online backup, online mind map viewer, and recommendations. Or you can use Docear as a local user with no data at all being submitted to Docear and no requirement for accepting any terms of service (just use Docear as you would use any other GPL desktop software).
After three demo-papers were accepted for JCDL 2013, we just received notice that another three posters were accepted for presentation at TPDL 2013 on Malta in September 2013. They cover some novel aspects of recommender systems relating to re-showing recommendations multiple times, considering user demographics when evaluating recommender systems, and investigating the effect of labelling recommendations. However, you can read the papers yourself, as we publish them as pre-print:
Paper 1: The Impact of Users’ Demographics (Age and Gender) and other Characteristics on Evaluating Recommender Systems (Download PDF | Doc)
In this paper we show the importance of considering demographics and other user characteristics when evaluating (research paper) recommender systems. We analyzed 37,572 recommendations delivered to 1,028 users and found that elderly users clicked more often on recommendations than younger ones. For instance, users with an age between 20 and 24 achieved click-through rates (CTR) of 2.73% on average while CTR for users between 50 and 54 was 9.26%. Gender only had a marginal impact (CTR males 6.88%; females 6.67%) but other user characteristics such as whether a user was registered (CTR: 6.95%) or not (4.97%) had a strong impact. Due to the results we argue that future research articles on recommender systems should report demographic data to make results better comparable.
Three of our submissions to the ACM/IEEE Joint Conference on Digital Libraries (JCDL) were accepted. They relate to recommender systems, reference management, and pdf metadata extraction:
In this demo-paper we introduce Docear4Word. Docear4Word enables researchers to insert and format their references and bibliographies in Microsoft Word, based on BibTeX and the Citation Style Language (CSL). Docear4Word features over 1,700 citation styles (Harvard, IEEE, ACM, etc.), is published as open source tool on http://docear.org, and runs with Microsoft Word 2002 and later on Windows XP and later. Docear4Word is similar to the MS-Word add-ons that reference managers like Endnote, Zotero, or Citavi offer with the difference that it is being developed to work with the de-facto standard BibTeX and hence to work with almost any reference manager.
Docear4Word 1.1 is available for download and it offers two new features that will improve your work with references in Microsoft Word a lot. Actually, we added two new elements to the “Add References” dialog.
The first one is a “Docear->Docear4Word” button. It’s intended for adding several references at once when you have multiple BibTeX keys in our clipboard. And here is how it works: Most references managers (e.g. JabRef and Docear) allow you to select several reference entries from the database and copy their BibTeX keys to the clipboard. That means you have a string like “Cohen05,ritchie2008,Eto12” in your clipboard. Now, when you press the “Docear->Docear4Word” button, Docear4Word will automatically get that string from the clipboard, identify the BibTeX keys and select the references belonging to the keys. This will make inserting several references at once much easier.
Docear’s PDF Inspector had a little bug: titles from PDFs were only extracted over the first two lines. That means when you had a PDF whose title expanded over three or more lines, only a part of the title was extracted. Read more…
Docear4Word 1.02 is available for download. It fixes one bug that occurred when adding several references at once. For instance, when you inserted three references [1-3], added another three references [4-6] and then added all six references again you would Read more…
We are glad to announce that we, again, will offer a paid internship in cooperation with the German Academic Exchange service (DAAD). If you are an undergraduate student, interested in software engineering or statistics, and coming from the Greece, Italy, Portugal, or Spain, get yourself started and do an 8-12 weeks internship in summer or autumn 2013, fully paid. And if you are not applicable to apply for the internship – please tell your friends to apply! 🙂
Your project
Your research question to answer will be “How to provide (better) research paper recommendations to our users?”. As such, it will be your task to support the Docear team in researching how the interests of Docear’s users can be identified from the users’ mind maps and how these interests can be matched with interesting items to recommend. You will do literature research, create new ideas, analyze user data, and implement new recommendation approaches in JAVA. Of course, you don’t have to do all of this alone – you will be closely cooperating with the Docear team. Your work will be integrated into Docear and used by thousands of researchers around the world. If your work is outstanding, we will write a research paper with you.
Requirements
You should have a profound knowledge of the programming language JAVA. Knowledge in statistics, machine learning, other programming languages (especially C/++ or Python) and/or MySQL, neo4j, Hibernate, Jersey, REST Web Services, Tomcat, and Apache is a plus, but not a requirement. Of course, we would appreciate if you spoke German but it would be no problem if you only spoke English. We would prefer, if you apply for a long internship (12 weeks) but you can also apply for a shorter internship. If you are interested in combining your internship with writing a Bachelor thesis, please let us know in advance (this would be highly welcome). You can start at any date you want in summer or autumn 2013.
Important: If you don’t want to program but have profound knowledge in statistics you are also very welcome to apply. In this case you will support us evaluating how good our current recommender system is, and you will help us generating ideas for improvements. Please indicate in your application clearly that you are not interested in software development but in statistics.
The 9th Beta of Docear is available for free download. It contains no new features but several bug fixes and feature improvements. One improvement includes the removal of line breaks in imported annotations. So far, when you highlighted text over several lines in a PDF, Docear imported the lines breaks of the highlighted text which sometimes caused a not so nice layout. In the new Docear you can do a right click on a node that links an annotation and select “PDF->Remove lines breaks from annotation”. Bug fixes include a fix for the bug that the Adobe Acrobat Professional PDF Viewer wasn’t recognized under MacOS. On Linux the splash screen does not hide any more the setup screen on the very first start of Docear. For a detailed list of all changes see the following change log. And keep in mind – we are always looking forward for your feedback and don’t forget to have a look at the new preview of Docear’s Online Viewer.
Feature Enhancements:
A few month ago we announced to develop an add-on allowing researchers to collaborate on the same data. Well, we haven’t finished this completely yet but the first step is done. In the past few month we developed an online viewer that allows you to view your backuped mind maps in your web browser. Right now the viewer is only capable of displaying very small mind maps but it’s our highest priority to improve the performance so you can also view larger mind maps. In addition, the online viewer displays only the basic elements such as nodes and edges. Other features such as attributes are not yet displayed but, again, we are working on it :-). Of course, this is just the beginning. As a next step Docear Online will allow you to edit your mind maps online. The following step will enable you to work simultaneously with different researchers on the same data either in your browser or with your desktop version of Docear.
To have a look at the preview version of Docear Online, go to https://my.docear.org and log in with your Docear user name and password. You will be able to view all those mind maps that have been created with the desktop version of Docear and that have been backuped with it. Simply select your mind map in the upper-right corner in the menu. To activate the backup function start Docear Desktop and open Tools->Preferences->Online Services->Manage Docear Service Settings.
We are experiencing a very high server load due to several reasons (many people are using our services, we are doing some extensive research analyses, etc.). Therefore we decided to deactivate the metadata retrieval and recommendations for a while, hopefully Read more…
Today we released Docear4Word 1.01. The add-on for Microsoft Word allows you to manage your Docear references (and any other references stored as BibTeX) directly within Microsoft Word. Version 1.01 includes the following changes. Download it here!
A while ago we released Docear with automatic metadata extraction. Several users asked us if this functionality was also available as stand-alone tool. Therefore, we decided to create a little tool which we name “Docear’s PDF Inspector”. Docear’s PDF Inspector extracts titles from PDF files not from the PDF’s metadata but from its full-text. More precisely, Docear’s PDF Inspector extracts the full-text of the first page of a PDF and looks for the largest text in the upper third of that page. This text is returned as title. Of course, this does not always deliver the correct title (e.g. sometimes the journal name is formatted in a larger font size than an article’s title) but in about 70% you will get the correct one. The main features of Docear’s PDF Inspector are
Just a quick notice: Since a while Docear can automatically retrieve bibliographic information for PDF files. So far, only 15 requests per day per users were possible. We increased this limit to 50. If you have not yet tried this Read more…
As a Docear user you probably did some research before you decided to use Docear and maybe you stumbled upon the reference manager Mendeley. Mendeley definitely has some nice features and made it to one of the top reference management tools in the past few years (besides the fact that they don’t use mind maps for literature management, the main reason I wouldn’t use Mendeley is the fact that they store the annotations you make in PDFs in a proprietary format — this locks you in to Mendeley and makes it really hard/impossible to switch to another tool). Two days ago Techcrunch reported that the well known publisher Elsevier takes an interest in buying Mendeley for presumably 100.000.000 US$. That’s right: 100 Million US$. Considering that Mendeley is supposed to have 2 Million users that would be 50$ per user (and I don’t know if the 2 Million users are really active users). As far as I remember, the shareholders of Facebook payed about 100 Dollars per user when Facebook shares were first available at the stock market. Not bad :-).
What do you think? Is Mendeley worth 100 Million Dollar? Is it a smart move from Elsevier to buy Mendeley? And what are the consequences for Mendeley’s users since Elsevier is known for a very harsh publishing policy which lead to a boycott of Elsevier and lots of criticism by many academics).
Docear4Word is an add-on that allows managing your Docear references in Microsoft Word. It uses the citation style language (CSL), an open XML-based language to describe the formatting of citations and bibliographies. Not only Docear is using CSL but also other reference managers such as Zotero (who initiated the development of CSL) and Mendeley and they are all contributing their styles – this is why there are more than 2,000 citation styles you can use with Docear4Word, and the other reference managers. However, sometimes the citation style you need is not in the citation style repository and up to now it was quite challenging to create a new style (or edit an existing one).
During the past months, the Columbia University Libraries, Alfred P. Sloan Foundation, and Mendeley developed a WYSIWYG editor for citation styles (What You See Is What You Get). This editor makes it easier than ever to edit existing styles and create new ones. So, if you are missing a citation style for a journal or conference you are submitting a paper to, well… create it and send a big thank you to the three organizations making this possible! :-).