It is important to run next scripts from root directory!
-
You can use our data or make sure that everything works - run pagerank_preview.py two times with different flag PERSONALIZATION_MODE = True and after PERSONALIZATION_MODE = False, which is placed at the beginning of the file. Such you will get personalized_result_domain_ranks.json and result_domain_ranks.json, which we use for testing correctness of personalization.
-
For testing correctness of ranks of our Fast Pagerank you can use our data or run it by yourself. For this run src/get_methods_results.py to get comparison_domains_ranks.csv, which we will test soon.
-
Now it is the most interesting part -- testing. Run one by one cells in the Benchmarking_and_Testing.ipynb to check correctness of the results and visualize data.
-
To look at our results on the real data, parsed from the Internet, execute PageRank_Real_Data_Values_Comparison.ipynb .
-
we understand that the best way to test our work is to use other well-developed libraries and compare our results on the same data. This idea we have found in this repo . The author proposed good way of testing on random graphs, time measuring and found out the best libraries in python, which can calculate Pagerank
-
Comparing calculated values for the same domains with our Fast Pagerank and other libraries
-
Comparing ranks for the same domains
-
Comparing speed for the same domains
