Text Summarization using Knowledge Graphs
Build instruction (maybe move this to separate file later)
- Clone git repo to your PC.
- Include all the external jar files required.
- Specifically remove jgrapht.jar, stanford-openie-{models,only-models,src,}.jar if included.
- Check all filepaths before building.
How to parse the raw data:
- Set the environment variable PROJECT_HOME to root directory of the project. e.g. PROJECT_HOME=/home/user/workspace/TextSummarization
- Create folder named 'data' inside PROJECT_HOME.
- Download and extract the CNN 'stories' from the link https://cs.nyu.edu/~kcho/DMQA/ to 'data' folder.
- Run Parser.py, will take couple of minutes to finish. It'll store the parsed files in the 'data/parsed' directory. Each parsed file has two lines. First line is the header (summary) and second is text. File name can be used as ID for each (summary,text) pair. Directory structure should look like this
PROJECT_HOME
.
├── data
│ ├── parsed
│ │ ├──summary── 0001d1afc246a7964130f43ae940af6bc6c57f01.story
│ │ └──text── 0001d1afc246a7964130f43ae940af6bc6c57f01.story
│ ├── output── 0001d1afc246a7964130f43ae940af6bc6c57f01.story
│ └── stories
│ └── 0001d1afc246a7964130f43ae940af6bc6c57f01.story
├── graphGeneration.py
├── PageRank.java
├── Parser.java
├── README.md
└── textSummarization.java
develop