OntLRB Benchmark

The Linear Road Benchmark LRB is a well documented test suit for relational data stream systems in a configurable traffic scenario. This work describes an extension of LRB, called OntLRB, which aims at making the classical LRB available for the semantically enhanced stream-engines community. OntLRB extends LRB with additional benchmarking aspects regarding the semantics of the data, in particular aspects related to inference and, further, reasoning in the ontology-based data access paradigm (OBDA).

Preparation for Exareme

We provide a proof-of-concept implementation of OntLRB for the distributed data stream management system ExaStream.

Open the Exareme terminal

python exareme/lib/madis/src/mterm.py my.db

Import the LRB data

Use the LRB data generator to create historical tolls and cardatapoints. To skip this step, Uppala University provides these files for one express way here. Secondly, use the script makeAllSeg.sh to generate segment data for any number of expressways. The file AllSeg.csv contains data for up to two expressways by now. Thirdly, adjust the queries found in lrb.sql in the folder LRB-for-ExaSteam to your environment. Update paths to point to your car data files. Run the LRB queries in the Madis/Exareme terminal.

Run the classic Linear Road Benchmark

OUTPUT file:bal.csv delimiter:, select 2, Time, cast(Emit AS int) AS Emit, QID, Bal, 0 AS ResultTime
FROM (lrb start:0 end:10784 SELECT * FROM AccBalOutStr);

OUTPUT file:accalert.csv delimiter:, SELECT 1, Time, Emit, VID, AccSeg FROM (lrb start:0 end:10784
SELECT * FROM AccNotifyStr);

OUTPUT file:tollalert.csv delimiter:, SELECT 0, VID, Time, Emit, AvgSpd as Spd, Toll FROM
(lrb start:0 end:10784 SELECT * FROM TollStr);

The generated output files can be validated with the validator from LRB.

Run an ontology extended benchmark

Unfold a query from STARQL to ExaStream. For example:

StarqlToExaStream.jar featureQuery.txt ExaStreamResult.txt
Ontologies/featuresWidth4.owl Mappings/featuresWidth4.obda

Run the queries found in ExaStreamResult.txt in the Madis/Exareme terminal. On successful execution, run the actual benchmark with such a command:

OUTPUT file:featuresWidth4.csv delimiter:, SELECT Time, ifnull(Emit-Time,0) AS Delay,
NumCars FROM (lrb start:0 end:10784 SELECT _vid AS VID, wid AS Time,
Count(DISTINCT _vid) AS NumCars FROM FinalTolls_having);

The file featuresWidth4.csv is filled with tuples of the form (time, delay, # of toll notifications). This data can be used for evaluation.