Using Tesseract in Docling: Language Specification, Binarization, and RAG Limitations

[abdouDeepLearning]

I am currently exploring the use of Docling for OCR tasks and have the following questions:

1. How can I use Tesseract within Docling while specifying the language(s) for OCR? Is there a way to configure it for multilingual documents?
2. Does Docling support image preprocessing with custom binarization techniques, such as Otsu's method or adaptive thresholding, to enhance OCR accuracy?
3. While Retrieval-Augmented Generation (RAG) works well for academic search, I’ve noticed limitations in professional document analysis, especially with chunking techniques that risk losing context. Does Docling support or plan to integrate Contextual Document Embeddings to address this issue?