Similarity search and deduplication at scale
1–10 of 17 posts
Re: Similarity search and deduplication at scale
#2I have been working on an entity matching solution for two years now, and I have decided to write down some of the learning I picked up along the way. Turns out there are too many relevant details to cover in a single post, so I will cover the topic in multiple parts.
This first part is the high-level introduction, useful for project planning and architecture decisions that need to be made early in the development process. Any feedback is welcome, along with wishes for the follow-up parts if you have something specific that you would like to be covered.
Re: Similarity search and deduplication at scale
#3I would like to know if any of these techniques could be used for identifying articles that are either copies of each other, or near-copies, or different articles on the same story.
Re: Similarity search and deduplication at scale
#4I would like to know if any of these techniques could be used for identifying articles that are either copies of each other, or near-copies, or different articles on the same story.
You should be able to use embeddings for this (sort by the cosine similarity). eg OpenAI has an off the shelf offering: https://beta.openai.com/docs/guides/embeddings/what-are-embe...
We used something similar to build a “similar articles” feature & it gave us de-duplication essentially for free.
Re: Similarity search and deduplication at scale
#5I would like to know if any of these techniques could be used for identifying articles that are either copies of each other, or near-copies, or different articles on the same story.
The easiest and likely most effective method may be to compute vector embeddings using a sentence transformer model, and find nearest neighbors among these vectors for all articles in the set. The distance between the nearest vectors will give you a degree of similarity between the articles. You'll need to figure out some thresholds on these distances to figure out what are near copies vs different articles on the same story. There are efficient methods to find approximate nearest neighbors among a large set of these vectors - available as both OSS and SaaS. Faiss [1], ScaNN [2], and Pinecone [3] are some examples.
This is one of the methods mentioned in the article. I don't have implementation experience with the other string distance measures in the article (under "normalized string" in the table), except for Q-grams. Compared to the above method Q-grams don't scale as well and are not as robust because it doesn't encapsulate an understanding of the semantics of the text.
[1] github.com/facebookresearch/faiss
[2] github.com/google-research/google-research/tree/master/scann
[3] www.pinecone.io
Re: Similarity search and deduplication at scale
#6I have been working on an entity matching solution for two years now, and I have decided to write down some of the learning I picked up along the way. Turns out there are too many relevant details to cover in a single post, so I will cover the topic in multiple parts. This first part is the high-level introduction, useful for project planning and architecture decisions that need to be made early in the development pr…
Thank you for this writeup. Having done some work on deduplication/matching systems, my experience (as with many things in data science) is that there are a lot of things consider and there is no single best solution. Hopefully you are able to keep up with this series, because I think it will be very helpful to many people.
Re: Similarity search and deduplication at scale
#7I would like to know if any of these techniques could be used for identifying articles that are either copies of each other, or near-copies, or different articles on the same story.
Exact and near duplicate articles should have similar or identical word frequency distributions. Maybe that can be used as a blocking criterion somehow. Although it might not be any faster to compare word frequency distributions than to compare dense low-dimensional embeddings.
Re: Similarity search and deduplication at scale
#8I would like to know if any of these techniques could be used for identifying articles that are either copies of each other, or near-copies, or different articles on the same story.
The easiest and likely most effective method may be to compute vector embeddings using a sentence transformer model, and find nearest neighbors among these vectors for all articles in the set. The distance between the nearest vectors will give you a degree of similarity between the articles. You'll need to figure out some thresholds on these distances to figure out what are near copies vs different articles on the sa…
If looking for exact or near-exact duplicates, a transformer seems like it's probably overkill. Maybe it's not bad if you already have one that you can use for inference in the database, but I suspect that something as simple as Fasttext would do the job. A transformer would probably be more useful if you want to catch things like replacing words with synonyms out of a thesaurus.
Re: Similarity search and deduplication at scale
#9I'm surprised to see that ML-based semantic search is barely touched on in this article. There's a strong focus on entity matching, but an arguably more powerful way to conduct similarity search is to leverage embedding vectors from trained models.
A great upside to this approach is that it works for a variety of different types of unstructured data (images, video, molecular structures, geospatial data, etc), not just text. The rise of multimodal models such as CLIP (https://openai.com/blog/clip) makes this even more relevant today. Combine it with a vector database such as Milvus (https://milvus.io) and you'll be able to do this at scale with very minimal effort.
Re: Similarity search and deduplication at scale
#10I would like to know if any of these techniques could be used for identifying articles that are either copies of each other, or near-copies, or different articles on the same story.
Yes, see these examples, which frame this as plagiarism detection but effectively do the same thing:
- https://dzone.com/articles/build-a-plagiarism-checker-using-...