wineseλrch — sample semantic data application for the Software Languages Team and Institute for Web Science of the University of Koblenz-Landau

For building, you'll need make, Gradle and yarn.

1. Have Semserv built and running.

2. Build Semantics4J and copy or symlink the resulting semantics.jar into the backend directory.

3. Run make and let the application build and start the server.

4. Visit http://localhost:3000/.

This is a single-page web application. It uses Semantics4J and Spark on the backend and Aurelia on the frontend.

The application provides the following features, each of them utilizing a different kind of data source.

The primary feature of wineseλrch is searching wines of course. It uses the Wine Ontology as its data source and Semantics4J for its primary implementation.

The user can specify a set of criteria to search by and receives a filtered set of results. The query simply builds a union out of each specified category, such as body or color, and then intersects all those unions.

By clicking on a wine in the result list, additional details can be viewed, such as maker and region of origin.

For regions and wineries, additional information can be obtained by clicking on its name in the detail view of a wine. The information will be shown in a modal window.

The source for this is the REST API of Wikipedia. The application performs a search for the given topic and shows a summary of the best-matching result. While this is just a best-effort guess, if there exists a Wikipedia article for the given topic, it will usually be chosen correctly.

Wines can be given ratings and reviews. Users can “log in” (really just enter a name, actual logins are elided as this is just an example) and give each wine a rating and review. Ratings are based on stars, with 1 star being the wost and 5 stars being the best. The average star rating, rounded to the nearest half, is displayed in the detail view of each wine.

Ratings use a local SQLite relational database to store and query ratings and reviews.

This application is separated into independent backend and frontend parts. The backend is the web application server and the frontend is the web browser client, which communicate via JSON over a REST interface.

By default, running make will build the production version of the frontend and set it up to be served by the backend server. The folder where the JavaScript files are served from is timestamped for the sake of cache busting.

During development, it may be easier to run the two pieces separately, as the frontend takes a relatively long time to build. For that, run make in the backend and frontend directories separately and visit http://localhost:9000/ instead. This page will automatically reload when changes are made to the frontend. The backend server still needs to be recompiled manually when something is modified, but it builds much more quickly.

There is an alternate implementation in the git branch stardog using Stardog version 4.2.4, a knowledge graph server with a SPARQL interface. This interface is not recommended, as it's extremely, prohibitively slow for this application.

If you still want to try it, git checkout stardog, start the Stardog server with the --snarl option, load the Wine Ontology as a database called wine. Then and run make and visit http://localhost:3000/ as usual.

Copyright 2017 Carsten Hartenfels.

Licensed under the Apache License, Version 2.

• Semantics4J

• Semserv

• The Wine Ontology

• λ-DL

• Aurelia