This repository presents a detailed analysis of academic theses related to Dr. Babak Teimourpour using data from the IranDoc platform, a prominent Iranian research documentation center. The project is split into three parts, each focusing on a specific task: keyword analysis, co-word network construction, and community detection within a co-word network. The repository contains three Jupyter notebooks, each addressing different aspects of network analysis and community detection in academic research.
The IranDoc platform, formally known as پژوهشگاه علوم و فناوری اطلاعات ایران, was established in 1968 under the name "Iranian Documentation Center." IranDoc provides a repository of scientific and academic resources, including theses, dissertations, and research reports. It plays a central role in the management and dissemination of scientific information in Iran. IranDoc is home to one of the largest archives of Iranian theses and is widely used by researchers for retrieving academic content. This project uses data sourced from IranDoc to analyze and visualize keyword co-occurrences from theses related to Dr. Teimourpour's research.
Objective: This notebook focuses on extracting and analyzing keywords from theses supervised by Dr. Babak Teimourpour. The primary goal is to determine the main research topics of the professor based on these keywords.
- Data Source: A filtered dataset from IranDoc.
- Process:
- Fetching thesis data from IranDoc.
- Extracting keywords from the theses.
- Counting occurrences of each keyword.
- Visualizing the distribution of keywords in a histogram.
- Outcome: Identification of prominent research topics through keyword analysis.
Objective: This part involves constructing a co-word network from the keywords of theses related to Dr. Teimourpour and performing network analysis.
- Data Extraction: Keywords from the IranDoc dataset are used to build the co-word network.
- Network Analysis:
- Construction of a co-word network using igraph.
- Calculation of centrality measures such as degree centrality and betweenness centrality.
- Visualization of the network with node sizes proportional to their centrality values.
- Export to Gephi: Centralities are exported for further visualization in Gephi.
- Outcome: A comprehensive co-word network, along with an interpretation of key nodes and their significance in Dr. Teimourpour's research.
Objective: In this part, community detection techniques are applied to the co-word network to identify clusters of related keywords representing academic concepts.
- Methods:
- Louvain Method: Detects communities and visualizes them with distinct color coding. The modularity of the communities is reported.
- Spinglass Method: Another method of community detection, with similar visualization and modularity reporting.
- Comparison:
- Communities from both methods are compared using the Rand Index to evaluate similarity.
- Modularity scores are used to assess the quality of the detected communities.
- Outcome: Visualizations of communities, comparisons of community structures, and insights into how keywords are grouped within the co-word network.
- Python: For data extraction, network construction, and analysis.
- igraph: A Python library for network analysis.
- numpy: Used for matrix operations and data processing.
- Gephi: For advanced network visualization (exported data).
- Jupyter Notebooks: The analysis is structured across three notebooks for easy replication and understanding.
This repository provides a detailed study of academic keyword networks, from basic keyword frequency analysis to more complex network structures and community detection techniques. It offers insights into Dr. Babak Teimourpour's research areas and demonstrates how network science can be applied to academic data.
Feel free to explore each notebook to follow the analysis steps in detail.