Webdext

Webdext is a Javascript library for web data extraction (web scraping). Currently, it only supports data records extraction from a list page (a web page containing 2 or more data records).

In order to use it, you must run Webdext inside the web page context. There are 2 ways to do that:

1. Use it as browser extension (currently, I only implemented the Chrome extension)
2. Inject the script into the web page context using headless browser such as PhantomJS or Splash (currently, I only implemented the runner script for PhantomJS)

Installation and usage

1. Chrome Extension
2. PhantomJS script

Internals

1. Intelligent extraction algorithm is heavily based on AutoRM [1] and DAG-MTM [2] (not an exact implementation though).
2. XPath wrapper induction algorithm is based on [3].

[1]	Shengsheng Shi , Chengfei Liu, Yi Shen, Chunfeng Yuan, Yihua Huang. 2015. AutoRM: An effective approach for automatic Web data record mining. Knowledge-Based Systems, 89, 314–331. doi:10.1016/j.knosys.2015.07.012

[2]	Shengsheng Shi , Chengfei Liu, Chunfeng Yuan, Yihua Huang. 2014. Multi-feature and DAG-based multi-tree matching algorithm for automatic web data mining. Proceedings of International Joint Conferences on Web Intelligence and Intelligent Agent Technology, 739–755. doi:10.1109/WI-IAT.2014.24

[3]	Joachim Nielandt, Antoon Bronselaer, Guy de Tré. 2016. Predicate enrichment of aligned XPaths for wrapper induction. Expert Systems With Applications 51, 259–275. doi:10.1016/j.eswa.2015.12.040

Author

Sigit Dewanto, sigitdewanto11[at]yahoo[dot]co[dot]uk