MathWebSearch / mws

MathWebSearch Implementation
https://search.mathweb.org/
GNU General Public License v3.0
47 stars 12 forks source link

MathWebSearch Build Status

About

The MathWebSearch system (MWS) is a content-based search engine for mathematical formulae. It indexes MathML formulae, using a technique derived from automated theorem proving: Substitution Tree Indexing. The software is licensed under the GNU General Public License version 3.

Demos

can be found at http://search.mathweb.org, in particular, the arXiv demo is at http://ar5search.kwarc.info

Content

Compiling and Testing

Compiling of the source tree is automated via CMake. You can build the sources using the following command:

make

Binaries are built in the bin/ directory, while documentation in bin/docs. To select or de-select which components to compile, use:

make config

To run the tests, use:

make test

Finally, install the binaries using:

make install

Dependencies

To build this software, one needs:

The core MathWebSearch executables require:

The crawler executables require:

The documentation target requires:

The test target requires:

The config step requires:

To install all build, runtime and test dependencies on a sufficiently new Debian / Ubuntu machine you can use:

apt-get install g++ cmake make pkg-config libmicrohttpd12 libxml2 libleveldb1v5 \
  libsnappy1v5 libjson-c-dev libhtmlcxx3v5 libgnutlsxx28 libicu57 libcurl3-gnutls

To install all build, runtime and test dependencies on Mac OS, you can use homebrew:

# install all the depependencies
brew install gcc make cmake pkg-config libmicrohttpd libxml2 leveldb snappy json-c \
  htmlcxx icu4c gnutls netcat curl
# in order for icu4c to be found also run:
brew link --force icu4c

Makefile targets

Usage

To use the Crawler, one needs to start the crawler with the defined website, the count of sites to crawl and optionaly: the start of the crawling and links to be skipped when crawling.

Another way to generate harvests is via docs2harvest. This takes as arguments XHTML documents and crawls them, creating harvests files. To crawl a repository of XHTML documents, use:

find . -name *.xhtml | xargs -n 10 bin/docs2harvest -o /path/to/harvests

The executable mwsd starts the main MWS server. This takes as argument a harvest include path which is used to load document data, and a port which where the data is served via HTTP. It accepts HTTP POST requests with MWS Queries and returns MWS Answer Sets.

bin/mwsd -I <harvest include dir> -p <port number>

For additional options, see:

bin/mwsd --help
bin/docs2harvest --help
bin/mws-config help

To setup or remove mwsd as a global SysV service, use (as root):

mws-config create -p 9090 -i data/zbl zbldemo
mws-config enable zbldemo

This will deploy MathWebSearch to serve the ZBL demo harvests on port 9090. To monitor, start or stop the service, use

service mwsd_zbldemo [start|stop|status|...]

Output is logged to /var/log/mwsd_zbldemo.log. To serve different harvest paths, create your own configurations and deploy the service.

Usage with Docker

This repository contains a Dockerfile for using the MWS Daemon. It can be found as the mathwebsearch/mathwebsearch automated build on Docker Hub and used as follows:

docker run -v /path/to/harvests:/data/ -p 8080:8080 mathwebsearch/mathwebsearch

The image is configured to serve harvests from a /data/ volume on port 8080.

Copyright

The software in this project (binaries and sources) is released "as is", under the GNU Public License version 3. A copy of this license can be found in the root of this project, under the file name LICENSE.

Credits

Most of the code in the core repository was developed by Corneliu-Claudiu Prodescu, under the supervision of Prof. Michael Kohlhase. For a complete list of developers visit https://github.com/KWARC/mws/graphs/contributors

Contact

The easiest way to contact the developers is using the MathWebSearch mailing list.