Master Thesis

WebQSP-WD available at: https://public.ukp.informatik.tu-darmstadt.de/coling2018-graph-neural-networks-question-answering/WebQSP_WD_v1.zip

Scripts

Please before run make sure to set correct parameters inside the .env file and each script. You can run them from the project root directory.

Data Preparation

🛠 Prepare Training Dataset

./reproduce/data__01_prepare_train_dataset.sh

🛠 Prepare Test Dataset

./reproduce/data__02_prepare_test_dataset.sh

🛠 Fetch Wiki articles

./reproduce/data__03_fetch_wikipedia_articles.sh

srtk tools - preprocess, train, retrieve + evaluate

🛠 Preprocess

./reproduce/srtk__01_preprocess_data.sh

🛠 Train

./reproduce/srtk__02_train_retriever.sh

🛠 Retrieve

./reproduce/srtk__03_retrieve_subgraphs.sh

SubgraphRAG

🛠 Build labels map

./reproduce/sg__00_update_labels_map.sh

🛠 Construct subgraphs from triplets

./reproduce/sg__01_construct_subgraphs.sh

🛠 Generate query-focused summaries

./reproduce/sg__02_fuse_contexts.sh

🛠 Generate answers

./reproduce/sg__03_generate_precise_answers.sh

🛠 Evaluate

./reproduce/sg__04_eval_subgraphrag.sh

Baselines

🛠 Run naive RAG experiment

./reproduce/baselines__naiverag_experiment.sh

🛠 Build Lightrag index

./reproduce/baselines__01_build_lightrag_index.sh

🛠 Evaluate

./reproduce/baselines__02_lightrag_experiment.sh