The community member is asking for viable approaches to update an ElasticSearchVectorstore index with only the PDF files that have changed or been removed, without having to re-index all the files again, in order to save on embedding costs. Another community member provided an example for a similar issue when using a vector database integration, mentioning that the process is more complicated and requires persisting/maintaining extra files. The example is for ChromaDB, but the community member notes it should work for any vector store.
Are there any viable approches for this issue: When loading all the documents for an index using a PDF loader, how can I only update the index with PDF files that have changed or have been removed using the ElasticSearchVectorstore without re-indexing all the files again, thus saving embedding costs?
