About Sovoria
Sovoria is an independent search engine that crawls and indexes the web itself.
Goal
Sovoria is building an independent search engine with its own crawler and search index, without relying on external search providers or combining results from other search engines.
Building and maintaining an independent web index takes time, but it allows Sovoria to serve search results entirely from its own infrastructure.
Why
Having multiple independent search indexes is important for a diverse and resilient web. It reduces reliance on a small number of providers and allows different approaches to crawling, ranking, and search quality.
By maintaining its own index, Sovoria can evolve independently without depending on the systems or decisions of another search provider.
Project
Sovoria is developed in my spare time and is entirely self-funded.
One of the project's priorities is efficiency. The crawler, index, and search backend are built to run on low-end hardware so operating costs stay manageable as the project grows.
Efficiency also influences how the crawler operates. Rather than downloading everything it can, it tries to collect the information needed to build a useful search index while avoiding unnecessary traffic.
There are no advertisements, sponsored results, or external search providers involved. This allows the project to grow at its own pace while remaining independent.
Privacy
Sovoria uses only first-party cookies that are needed to protect forms and remember your selected theme, favicon preference, diagnostics settings, and search filters.
The search-filter cookie stores only your selected topics, site types, page types, and Safe Search setting. It does not store search queries. You can remove these settings at any time by opening Filters and selecting Clear all.
Sovoria does not use advertising cookies or cross-site tracking cookies.
When favicons are enabled, your browser requests each website's declared icon
directly, or
/favicon.ico if no icon was declared. This means the website hosting the icon
receives
the
normal request from your browser.
Anonymous search improvement diagnostics is enabled by default and can be disabled from the search actions. When enabled, searches that take at least one second are stored in a private administration database together with the search terms, timestamp, total duration, and timing breakdown. Clicks may also be counted in aggregate to help improve ranking accuracy and re-crawl priorities. This information is used to diagnose search performance and improve ranking. Disabling diagnostics prevents any of this from being stored.
Image-search thumbnails are requested directly from the websites that host them. Sovoria does not send the search page as a referrer, but the image host still receives the normal network request from your browser.
Contact
Questions about crawling, indexing, or your website are always welcome.
Please don't include sensitive information in your first email. If necessary, another communication channel can be arranged afterwards.
More information can be found on my personal page taucher.dev
Indexing & Ranking
Sovoria indexes publicly accessible web pages using its own crawler. It follows standard web
conventions such as robots.txt, crawl delays, and sitemap files.
It identifies itself as sovoria.eu, avoids unnecessary downloads, and slows
down automatically if a server appears busy or responds with rate limits.
Not every page that is crawled is added to the search index. Pages are evaluated for factors such as originality, structure, and overall quality. Pages that appear to be spam, automatically generated with little value, or consist mainly of duplicate content may be excluded from the index.
Search ranking considers a combination of signals, including page content, links, metadata, and how a page relates to a search query. The exact ranking methods continue to evolve as the project grows.
Open Source
Currently the Sovoria source code itself is not publicly available.
Sovoria is built with open-source software. The projects, datasets, and licenses used are listed in the open-source acknowledgements.