Are you guys using substring match for searching documents or vector based similarity techniques?

How do you think the "mentions" capability within the "company" section in the grapevine app work ? It's not like someone is tagging a company like they can tag a user, right !! Also, a substring match would not also work since the company name can be quite generic and then you could have irrelevant posts.. And keeping a human who is doing manual annotation also does not make sense because it's literally not scalable... What do you think ?

I have made a search for my project using elastic search in which i used synonyms and and fuzziness and basic matching. I want it to work as users feeling, what users want , can you guys help me how to implement it in elastic search, i tried its semantic search but couldn't get its ELSR model. Help me!!!!

Has anyone worked with ML problems where we need to match N elements from 1 dataset to M elements from another (N>M, or 1 m can have multiple n)? What business domains have such problems? What modelling approaches, feature engineering and ML models did you use?

Or might build in the future

I ran the simulations with two random vectors of length 100, and they were indeed atleast 45 degrees apart. But I can't seem to understand why? My approach: 1. Take two vectors of length 100. Randomly allocate values to each component. 2. Compute inverse cosine of ( a . b / |a| |b| ) The histogram is centered around 45 degrees. but why?

Hey folks, I have been working for the past few months on the centralised internal search tool for all the companies and I need your honest feedback and feedback. Problem: Today, every company uses "a ton" of tools- Notion, confluence, Drive, PDFs etc. and the information is scattered and hard to find. Solution: I (along with my co-founder) am working on a centralised search tool where you connect all of these tools as integrations and voila you find it. Plus, we are also adding a layer of asking questions, "What is our churn rate" Instead of finding a document and then navigating through it to find the answer, you will get the answer right there. Ask: Please give me your feedback/suggestions. Are you going to use it?

I'm building an order matching engine. But it's based on qty bought/sold (and not price). Idea is to fully match buy orders and sell orders (and not partially match) in order to settle 2 or more parties. I'm planning to create it on make.com or firebase functions (whichever is cheaper), using firestore as db. What's the best possible approach? PS: Open to answering questions for better context via comments or DM