hep_filter

hep_filter is python script that intercepts daily abstract announcements by arXiv.org, calculates a score of each abstract according to keywords from your scores file and than reorders abstracts so that the ones most interesting to you come out on top. Resulting email is then forwarded to you.

Download

Current version is hep_filter 0.6 [2007-08-24]. [download]

Installation

Firstly, download the hep_filter script somewhere and make it executable. Then configure your scores file as explained below. Path to this file should be set at the begining of the hep_filter script.

Note also that hep_filter mails resulting abstracts to you (address specified at the beginning of the script) using your local SMTP server "localhost". Edit hep_filter if you need to change this or if you need to specify login and password.

Finally, you need to configure your email so that arXiv.org mailings get redirected to hep_filter script. Ask your local email guru how to do this. Let me just say that I use procmail on Unix and the following recipes in my .procmailrc file:

 
# If already hep_filtered hep-daily goes to INBOX for reading
:0:
* ^From.*hep_filter*
INBOX

# Otherwise filter it by hep_filter
:0
* ^From.*no-reply@arXiv.org*
| /path_to/hep_filter

Scores file

Syntax of this file is simple. Each line is of the form: string -> score

 
Matrix theor% -> -3
orbifold -> -3
Witten -> +8
cosmological constant -> 1
revised version -> -5
Bell% inequalit% -> 3

All lines without "->" are ignored.
Percent sign is a wildcard matching anything until the end of the word. You cannot use it in the middle of the word.
Pattern matching is case insensitive.
Matched strings will be enclosed within *-signs in the abstracts. You can instruct your email reader to boldify or colour these strings.

Testing

To see how it works without reconfiguring your email setup just save one arXiv.org email to a file (preserving email headers!) and run a "hep_filter < thisfile". SMTP server such as sendmail still needs to be running.

Bugs

Patterns are recognized even in the middle of the word. So if you don't want to read about Lie algebras and Lie derivatives and score "Lie", script will also score words such as "implies".

[ webmaster ]

Last update: 2007-08-24