devrel
devrel copied to clipboard
meilisearch
Create the v1 launch demo
In your wildest dreams, what demo would be the most awesome to see on MeiliSearch? The biggest wow factor for you, even if you think it's not possible right now with the product we have. What is the ONE thing that would feel like there is a party in heaven made exclusively for your eyes when you check the demo of our new version... Be wild! Be creative! Think outside of the box!
Thinking about v1 launch, we should update the main demos to have something even more amazing and huge. Preferably something with a proper size of information per document and with one field being something visual (image or video...).
I would love to be able to search through the lyrics of the rolling stone magazine top 500 music album and having the autoplay on, but I am not sure we would have the right to do something like this 😢 Maybe if I connect with my Spotify account? 🤷
I would love to search in a massive dataset like the 115m+ songs from discogs.com. It would be great to show users that we can search, facet, and filter on a massive amount of data now, that this is no more a problem.
I also thought about being able to search in your whole Twitter history, liked, retweeted, or even just "seen" tweets. I know that it would be a little more complex to develop as we need the user's Twitter account.
I would like to have a search on fake data. Like 50M fake users for a CRM like, or search on GitHub users. Search on fake conversations, like a fake chat app.
It would be also so cool to have a search on Wikipedia (need massive machines), but only index a few things like the title, and the first paragraph.
- Create a Pokedex to search on all Pokemon! Quite fun to design, but not so many documents to manage.
- Search on Magic card game. I found a list!
- Search for every book present on Goodreads.
I would love to search in a massive dataset like the 115m+ songs from discogs.com. It would be great to show users that we can search, facet, and filter on a massive amount of data now, that this is no more a problem.
I also thought about being able to search in your whole Twitter history, liked, retweeted, or even just "seen" tweets. I know that it would be a little more complex to develop as we need the user's Twitter account.
Seconding this, I would love to see how Meilisearch could compete with Elasticsearch when searching through large datasets that contains billions of data. Would be great to have performance benchmarks and showcase about how fast indexing is compared to competitors. (Many articles say that one of the biggest con of Meilisearch is slow indexing on large datasets) Also adding a search suggestions would be great.
Four words: searchable database of memes 😄 Would be difficult since most memes are images, so would require OCR. But can you imagine being able to quickly find the best meme for any given situation (like Giphy/other GIF keyboards)?
Some random idea, search on Pokémon or a card game like D&D, It could be fun for the designer.
Big +1 to the D&D idea, most of that stuff is text already so it would be easy to make, and could actually be hugely helpful to a niche community. Database of D&D weapons/objects, for example.
I would love to search in a massive dataset like the 115m+ songs from discogs.com. It would be great to show users that we can search, facet, and filter on a massive amount of data now, that this is no more a problem. I also thought about being able to search in your whole Twitter history, liked, retweeted, or even just "seen" tweets. I know that it would be a little more complex to develop as we need the user's Twitter account.
Seconding this, I would love to see how Meilisearch could compete with Elasticsearch when searching through large datasets that contains billions of data. Would be great to have performance benchmarks and showcase about how fast indexing is compared to competitors. (Many articles say that one of the biggest con of Meilisearch is slow indexing on large datasets)
Totally agree as well. Additionally, it would be great to showcase in the same demo:
- Query suggestions (that improve over time with analytics)
- Sort the search results on the fly (e.g., by createdAt, alphabetically, popularity, relevance, etc.)
- With a distributed search system such that results appear instantaneous wherever you are in the world (100ms from keystroke to displayed results)
- If you select a song, it will show you "more songs like this"
- If you spell something incorrectly, "did you mean …"
- With an open-source version of Algolia's "Understand" feature that not only tokenizes but also parses the queries with NLP to produce super relevant results
- With a live dashboard that anyone can view with statistics about the demo (e.g., with searches, average search time, most popular queries, most recent searches, average click position on the results, etc.)
- With a few videos showing (in real-time) how to get started with MeiliSearch and update the engine. Proof that you can unlock all these features in just 5 minutes.
None of those are easy (or possible) with ANY open-source search engine that I found, so even just one with billions of pieces of data would be impressive. That would be wow.
If you're looking for demos that not only showcase real-world use cases but also ones that highlight the special features of MeiliSearch, create a demo that visually shows how MeiliSearch parses the query and sorts the results. That way, people can visualize the tokenizer and the ranking rules on the fly.
By no means do I expect any of this to happen before v1, just thinking big.
