Weaviate
.svg){kind=logo} | |
 | |
| Developer(s) | SeMI Technologies BV Bob van Luijt |
|---|---|
| Initial release | March 30, 2016[1] |
| Stable release | |
| Repository | github |
| Written in | Go |
| Engine | |
| Operating system | Cross-platform |
| Type | vector search engine |
| License | Various (open-core model), e.g. BSD 3.0, Weaviate License (proprietary) |
Search Weaviate on Amazon.
Weaviate is a vector search engine[5] and vector database[6] with full CRUD support[7] based on approximate nearest neighbor algorithms (such as Hierarchical Navigable Small World graphs[8]).
Weaviate uses a graph-like data model in which all nodes are represented as an n-dimensional space vector in a vector space.[9] This representation can be set manually or through modules[10] (e.g., based on Transformers[11], image vectorization, contextionaries, etc). This allows the user to conduct similarity searches and automated classifications on large datasets. Most use cases are dealing with unstructured data where the machine learning models that vectorized the data are used to automatically predict the right relations between data objects.[12]
History[edit]
During an interview on FLOSS Weekly[13] and in ZDNet[5], Bob van Luijt explained that in 2016, he was inspired by the combination of the news that Google Search changed their search algorithms to vectorization[14] (i.e., RankBrain) and because of the growing popularity of word embeddings. He believed that it should be possible to create a software solution that would make this technology available to others as well.[15]
The initial concept was to enrich existing graph databases[16] with vector representations but when existing database technology seemed insufficient to handle large amounts of vectorized data-objects, Weaviate changed into being a vector database on its own specifically focussing on searching through large amounts of vectorized data objects.[17]
References[edit]
- ↑ "Weaviate initial commit". Github. 2016.
- ↑ "List of Weaviate releases". Github.
- ↑ "Weaviate on DB-Engines". db-engines.com.
- ↑ "Weaviate v1.0.4 on Github". Github.
- ↑ 5.0 5.1 Anadiotis, George. "Weaviate is an open-source search engine powered by ML, vectors, graphs, and GraphQL". ZDNet. ZDNet. Retrieved 9 April 2021.
- ↑ van Genuchten M, Hatton L, Verhagen M, van Luijt BM (2020). "Bringing Semantic Knowledge Graph Technology to Your Data". IEEE Software. 37 (2): 89–94. doi:10.1109/MS.2019.2957526. Unknown parameter
|s2cid=ignored (help) - ↑ Dilocker, Etienne. "Weaviate, an ANN Database with CRUD support". DB-engines. DB-engines. Retrieved 2 February 2021.
- ↑ Malkov, Yury; Yashunin, Dmitry (2016). "Efficient and robust approximate nearest neighbor search using Hierarchical Navigable Small World graphs". arXiv:1603.09320 [cs.DS].
- ↑ "Weaviate software documentation". SeMI Technologies.
- ↑ "Weaviate modules". SeMI Technologies.
- ↑ "A sub-50ms neural search with DistilBERT and Weaviate". Towards Data Science. 10 February 2021.
- ↑ Kwakernaak M (2020). "Interview with Technology Startup CEO Bob van Luijt on Value Creation in the Digital Age". Journal of Creating Value. 6 (2): 208–216. doi:10.1177/2394964320968996. Unknown parameter
|s2cid=ignored (help) - ↑ FLOSS Weekly episode list 2020
- ↑ Clark, Jack. "Google Turning Its Lucrative Web Search Over to AI Machines". Bloomberg Business. Bloomberg. Retrieved 28 October 2015.
- ↑ "Weaviate on FLOSS Weekly". FLOSS Weekly.
- ↑ Dilocker, Etienne. "Weaviate on FOSDEM 2019". FOSDEM.
- ↑ van Luijt, Bob. "Weaviate on FOSDEM 2020". FOSDEM.
External links[edit]
This article "Weaviate" is from Wikipedia. The list of its authors can be seen in its historical and/or the page Edithistory:Weaviate. Articles copied from Draft Namespace on Wikipedia could be seen on the Draft Namespace of Wikipedia and not main one.
