About AMSearch
AMSearch is a web-based application prototype developed as a search engine system for text documents in the Sundanese language. AMSearch supports semantic search and morphological word analysis in Sundanese using a rule-based approach and document vector representation.
Key Features of AMSearch
Document Search (IR System)
- Uses a Transformer embedding model (BERT-based).
- Search based on queries in Sundanese.
- Displays relevant document results from the Sundanese dataset.
Innovative Contributions
- A new stemming algorithm for Sundanese.
- A specialized IR dataset for Sundanese, which is rarely available (AMSunda Dataset).
- AMSearch website as a proof-of-concept for a local transformer-based IR system.
- Efforts to preserve and digitize the Sundanese language through NLP and IR.