contact-andy/Amharic_DBpedia_Template_Mapping_Automation

# Automated Mapping of Wikipedia Templates to DBpedia The process of mapping Wikipedia templates into DBpedia is often repetitive and labor-intensive. Each template, along with its attributes, needs to be identified, documented, and transformed into a structured mapping format. As the number of templates grows, manually repeating these steps becomes inefficient, time-consuming, and prone to inconsistencies. Automation is therefore essential to accelerate the mapping workflow, ensure consistency across templates, and support large-scale extraction efforts. ## 1. Why Automation? The process of creating Wikipedia templates and their corresponding DBpedia mappings is often repetitive, time-consuming, and error-prone if done manually. Each template contains multiple attributes (properties), which must be identified, mapped to DBpedia ontology, and kept consistent across different languages. Automating this process helps to: - Save time and effort by avoiding repetitive manual mapping. - Ensure consistency in attribute translation and ontology alignment. - Reduce errors when dealing with large numbers of templates. - Scale efficiently to support new templates, properties, and languages. - Enable reproducibility so that other contributors can repeat the process with the same steps. ## 2. Objectives of Automation The main objectives of automation in the Amharic DBpedia extraction are: - Automatically extract attributes (properties) from Wikipedia infobox templates. - Translate these attributes into Amharic using machine translation (e.g., NLLB or Gemini API). - Generate DBpedia-compliant mapping files with minimal manual intervention. - Build a pipeline that can be reused for future templates and extended to other Ethiopian languages. ## 3. Prerequisites Before running the automated extraction pipeline, the following are required: ### Technical Setup - Python environment with required libraries (`requests`, `json`, `re`, `transformers` if using NLLB). - Access to …