Home
amableDB is a fast (nosql) database which is specially designed for finding similar float-vectors. For that it depends on the amazing NMSLIB library. All CRUD-Operations will be supported via simple HTTP-Requests (GET and POST). So you can use it in any programming language you like!
For my project Hive Discover I had to do some KNN-Search over Post-Data. A neural network analyzed the text-body of different posts. As a result, I got 46 different values from 0 - 1 which represents the 46 categories. So to find similar posts I had to find similar vectors and I looked for a library for this kind of job. I ended up using the NMSLIB-Implementation of k-Nearest-Neighbor. I found out that elastic-search supports search operations but it required way to much memory. So I thought: Can this not be lightweight? Also elastic-search does not support directly the knn-Search.
That is why I created the amableDB project. It should help people who want to find quick and easy similar vectors with different search algorithms (currently only the knn-Search is available; later will be more). Also machine learning algorithms like SVC or Linear Regression are planned which should be easy to perform on collections.