Article_Graph

Article_Graph is a tool that extracts and enriches information from a set of academic papers and journals.

It makes use of advanced and powerful machine learning tools to extract as much information as possible. Also, it uses Grobid to extract all the relevant information about the papers.

The final output of this experiment is a RDF Graph that includes all the extracted and reconciled information about the papers and their relations.

A simple application is also available to visualize and interact with the KG.

Requirements

Python >= 3.11 is required for running the experiments.

Grobid is required for the first step of the pipeline, you can follow the installation instructions here.

PDF_ArticleAnalyzer is required to interact with the Grobid service, you can follow the installation instructions here.

Running the Application with Docker with the KG in a Remote Server

If you want to try the application with the pregenerated graph under the rdf directory, here you will find all the instructions necessary for running it.

1. Clone the repository:

git clone https://github.com/JorgeMIng/Article_Graph
cd Article_Graph

2. Build the Docker image:

docker build -t graph_tool docker

3. Run the image:

docker run -p 8501:8501 graph_tool

By default, the KG generated in the examples/article_graph.ipynb is loaded in a remote server http://yordi111nas.synology.me:3030/articles/query. The graph is also available under the rdf directory.

Running the Application from Source with the KG in a Remote Server

If you want to try the application with the pregenerated graph under the rdf directory, here you will find all the instructions necessary for running it.

1. Clone the repository:

git clone https://github.com/JorgeMIng/Article_Graph
cd Article_Graph

2. Create a Python environment (conda is recommended):

conda create -n article-graph-3.11 python=3.11
conda activate article-graph-3.11

3. Install all the dependencies:

pip install -r requirements_app.txt

4. Execute the application:

python Start.py

By default, the KG generated in the examples/article_graph.ipynb is loaded in a remote server http://yordi111nas.synology.me:3030/articles/query

Running the Application with a custom KG Hosted Locally

If you want to try the application with another graph generated locally, here you will find all the instructions necessary for running it.

1. Clone the repository:

git clone https://github.com/JorgeMIng/Article_Graph
cd Article_Graph

2. Create a Python environment (conda is recommended):

conda create -n article-graph-3.11 python=3.11
conda activate article-graph-3.11

3. Install all the dependencies:

pip install -r requirements_app.txt

4. Host the KG in Jena Fuseki with Docker:

docker run -p 3030:3030 stain/jena-fuseki

5. Execute the application:

python Start.py

6. Go to the Settings section and configure the remote server.

Running the Experiments

If you want to reproduce the experiments by yourself, here you will find all the instructions necessary for running them.

1. Clone the repository:

git clone https://github.com/JorgeMIng/Article_Graph
cd Article_Graph

2. Create a Python environment (conda is recommended):

conda create -n article-graph-3.11 python=3.11
conda activate article-graph-3.11

3. Install all the dependencies:

pip install -r requirements.txt

4. Run the example notebook at examples/article_graph.ipynb

Examples

• Full KG Generation: examples/article_graph.ipynb
• Similarity Analysis: examples/examples_similarity.ipynb
• Topic Modeling: examples/topic_modeling.ipynb
• NER Analysis and Project extracton: ner/extract_element.ipynb

License

Please refer to the LICENSE file.

Authors

• Jorge Martín Izquierdo
• Gloria Cumia Espinosa de los Monteros
• Marco Ciccalè Baztán