The application crawles Wiki pages, analyses the content using simple NLP methods and generates data in graph database.
Currently, the application analyze only some Polish Wiki pages - this is the only source of data so far. Application supports two database outputs: Neo4j and OrientDB.
The goals of the project:
- Build graph database of people connected via information available on the network.
- Learn and analyze data using available graph algorithms.
- Compare the performance between different graph databases.
Here are some performance results achieved during importing data:
Neo4j 3.3.3
Operation | Average time | Min time | Max time |
---|---|---|---|
Creating persons | avg=130.437 ms | min=41.471 ms | max=306.532 ms |
Creating relations | avg= 40.747 ms | min= 0.455 ms | max=136.625 ms |
OrientDB 3.0 RC2
OrientDB with index for Person.name
Operation | Average time | Min time | Max time |
---|---|---|---|
Creating persons | avg=1.336 ms | min=1.132 ms | max=8.293 ms |
Creating relations | avg=90.948 ms | min=78.615 ms | max=127.330 ms |
OrientDB with index for Person.name, index on Wiki.in and .out
Operation | Average time | Min time | Max time |
---|---|---|---|
Creating persons | avg=1.283 ms | min=1.058 ms | max=3.654 ms |
Creating relations | avg=1.781 ms | min=0.773 ms | max=5.032 ms |
Machine specification: Windows 10, CPU: AMD FX-6300 6 cores 3.5GHz, RAM 8 GB