pcminer

PC-miner is a tool that analyzes publications and program committee service in the top conferences in Programming Languages and Software Engineering. Currently, the tool supports the following conferences: ASE, ECOOP, ESOP, FSE, ICSE, ISSTA, OOPSLA, PLDI, and POPL and it covers the years 1995-2014. Publication data was obtained from DBLP, and program committee data was scraped from various conference web sites.

I. Using PC-miner

To start using PC-miner, go into the "ui" subdirectory, and open the index.html file in your browser. This should bring up a window that looks like the following screenshot.

In the selector on the left is an alphabetical list of all authors. Clicking on an author will display his publication and committee activity. You can search (awkwardly) by entering some text in the search box. Entering "Jan" there will reduce the list to all authors whose name starts with the characters "Jan". When you select Jan Vitek's name from the resulting list, you'll see something that looks like the attached screenshot. The file screenshot2.jpg shows Jan's record after restricting the date range to 2009-2014 using the slider.

Note that clicking on a co-author's name will result in navigating to the record for that co-author.

The textbox on the right can be used to search for people who meet certain criteria. For example, you might be interested in finding all people who served on both the FSE and ICSE committees by entering:

committees("FSE")>0 && committees("ICSE")>0

which produces a long list of familiar names.

Or you can search for all people who published at least 5 papers at PLDI but who haven't served on the PLDI PC yet, using:

publications("PLDI")>4 && committees("PLDI") == 0

Or you could, e.g., search for people who published at least 6 papers in { OOPSLA, ECOOP }, but haven't served on the ECOOP committee yet:

publications("OOPSLA")+publications("ECOOP")>5 && committees("ECOOP") == 0

There are some other bells and whistles. You can uncheck the checkboxes at the top of the tool to filter out conferences, and limit the period you want to consider using the slider. To copy the names currently displayed in the selector, press the "Display selected names" button.

II. Adding conferences to PC-miner

The PC-miner generator tool is a Java application that you can build and run using Eclipse. To run PC-miner, right-click on the class pcminer.PCMiner, and select Run As->Java Application. No configuration should be necessary.

The input to the generator tool is provided in the data subdirectory, which contains one subdirectory for each conference (e.g., "PLDI"). Each conference-subdirectory, in turn, contains one subdirectory for each year (e.g., "1999", "2000", etc.) as well as a file "color" that specifies the color to be used for the conference in the generated UI (see here for a list of HTML color codes). Each year-subdirectory contains two files: a file that was copied from DBLP containing the DBLP info for a specific year of a specific conference (e.g., here is a link to the file for PLDI 2000) and a file specifying the program committee for that year (e.g., here is the file that was created for PLDI 2000).

Adding a conference involves the following steps:

1. create a new subdirectory in the "data" directory with a name matching the conference acronym (e.g., "PLDI")

2. create a file "color" inside of this directory. This file should contain a single line that provides a name or hexcode of your preferred color (e.g, "#3BB9FF") See http://html-color-codes.info/color-names/ for a list of colors.

3. inside of this new conference directory, create a subdirectory for each year for which data is available (e.g., "2000", "2001", "2002", etc.)

4. inside each year subdirectory, you need to provide two files

   • a file ".html" providing the HTML source code for that year of that conference as provided by DBLP.

   • a file "-pc.txt" providing the list of PC members.

     Important: The name of each PC member must match the name and middle initials as they are used in DBLP (otherwise, get duplicate entries will appear in the UI). Note that special characters in people's names need to be encoded (see here for a list of HTML character codes). For example, you'll need to encode a lowercase 'a' with an umlaut as '&auml' or '&#228'.

     By default, including a person's name in the committee-file means that they are assumed to be a PC member. For other roles (e.g., program chairs), names need to be prefixed as follows:

     • "P:" - program chair
     • "G:" - general chair
     • "E:" - ERC member
     • "C:" - conference chair

5. Once your files are all in place, you can generate the UI by running PCMiner.java. This will generate an updated version of the file ui/data.js. Opening ui/index.html in your browser will bring up a version of PC-miner that contains the conference you just added. Please contribute any data you added, so that others can benefit from your efforts.

III. Feedback, Suggestions, and Acknowledgements

PC-miner was developed by Frank Tip as a tool to help identify PC member candidates for the ISSTA'11 and PLDI'12 program committees. Since then, it has been used by several other PC chairs for conferences such as ISSTA, ICSE, FSE, PLDI, ASE, ECOOP, and OOPSLA. Feedback, suggestions, and improvements are most welcome, as is additional conference data. Feedback, contributions and suggestions from Giuseppe Castagna, Matt Dwyer, Cormac Flanagan, Richard Jones, Max Schaefer, Alex Orso, Jan Vitek, and Andreas Zeller are gratefully acknowledged.

The data made available in PC-miner contains information from the DBLP Bibliography Server which is made available under the ODC Attribution License.

(c) 2010-2013 Frank Tip