Tentris: A Tensor-based Triple Store

Tentris is a tensor-based triplestore that natively supports worst-case optimal joins.

This is the research version of Tentris. A commercial version is available at https://github.com/tentris/tentris.

Features

• SPARQL endpoint supporting SELECT, SELECT DISTINCT and ASK queries at /sparql and /stream. The WHERE block may contain basic graph patterns and OPTIONAL clauses. Queries must be well-designed Kaminski et al. (2018)
• Supports INSERT DATA/DELETE DATA updates that are synchronized with a reader-writer lock at /update.
• Supports streaming results through the /stream endpoint for large results.
• Tentris is a persistent triplestore that uses metall to manage its disk-based index. Note that previous versions were in-memory.

Download

We provide pre-built binaries in the releases that run on Linux machines with fairly recent (after 2015) x86_64 CPUs.

Running

Initialize Tentris' index with an empty graph with:

./tentris_loader -f <(echo "")

Or initialize it by loading a turtle or ttl file:

tentris_loader --file my_nt_file.nt

Start the Tentris endpoint at 127.0.0.1:9080/sparql:

tentris_server -p 9080

Run a query with :

curl -G \
--data-urlencode 'query=SELECT * WHERE { ?s ?p ?o . }' \
'127.0.0.1:9080/sparql'

Or update your data with:

curl -X POST \
-H 'Content-Type: application/sparql-update' \
--data 'INSERT DATA { <http://example.org/subject> <http://example.org/predicate> "object" . }' \
'127.0.0.1:9080/sparql'

The following endpoints are available:

• HTTP GET /sparql for normal SPARQL queries
• HTTP GET /stream for SPARQL queries with huge results
• HTTP POST /update for INSERT DATA and DELETE DATA updates

Build It

Tentris is known to build on Ubuntu 22.04 and newer. Building was tested with Clang 17 & 19. As C++ standard template library, only libstdc++11 (v13) was tested. For details refer to the Dockerfile or GitHub actions.

Research

• Alexander Bigerl, Liss Heidrich, Nikolaos Karalis and Axel-Cyrille Ngonga Ngomo (2025) Efficient Updates for Worst-Case Optimal Join Triple Stores. In: The Semantic Web – ISWC 2025 | website

• Alexander Bigerl, Lixi Conrads, Charlotte Behning, Muhammad Saleem and Axel-Cyrille Ngonga Ngomo (2022) Hashing the Hypertrie: Space- and Time-Efficient Indexing for SPARQL in Tensors. In: The Semantic Web – ISWC 2022 | website

• Alexander Bigerl, Felix Conrads, Charlotte Behning, Mohamed Ahmed Sherif, Muhammad Saleem and Axel-Cyrille Ngonga Ngomo (2020) Tentris – A Tensor-Based Triple Store. In: The Semantic Web – ISWC 2020 | website