Master Thesis on Building a Collaborative-Filtering techniques-based Personalized PageRank.

Please note that this project is only aimed at running on Linux-based systems.

Dependencies are the following:

• ElasticSearch
• Python 2.7.x
• Graph-Tool Python library
• Scrapy Python library
• chardet, elasticsearch, pymongo and bs4 Python modules
• Bash
• ZSH
• Multimarkdown (if you want to recompile the CPPR.md file in report/)

Description of folders:

• algolia contains an attempt to use Algolia.com service as a search engine backend, did not work.
• elastic_search contains the configuration of ElasticSearch together with its launching scripts.
• frontend contains the user study platform
• logs-processing contains all scripts related to processing the logs: Usage extraction, user similarity, etc.
• logs-processing/logs-enrichment contains the scripts related to adding the missing data to the logs.
• ppr-java contains the test code to use JUNG to compute PageRank, reading from a GraphML file.
• ppr contains all scripts related to computing PageRank and Personalized PageRank.
• presentation contains files integrated in the final presentation and the defense presentation (exported from GDocs format).
• report contains the Master Thesis (Masterarbeit) written work source code, together with its dependency files (images, latex input files).
• scripts contains misc scripts, mainly setup scripts. Shorthands to do procedures like installing Privoxy and Tor in a docker.
• user_study contains results of the user study results (JSON sessions) together with the script necessary to generate the measures/metrics results from the raw user study results.
• versionned_data contains versionned "data", results and binaries (spreadsheets) that needed to be versionned for safety and to be able to follow their changes.
• web_crawler contains the web crawler.`

The current folder, EXCEPT THE FOLDER logs-processing/logs-enrichment, is under the LGPLv3 license. The folder logs-processing/logs-enrichment is not to be redistributed without express consent of its author.