Federated queries

One major advantage of linked open data is that we can use not only the information stored within our graph, but also write queries linking to different knowledge bases like Wikidata or others that link to Wikidata items. This is possible because we have matched most of our values to existing Wikidata items.

Since we have location information like places of publication or narrative locations of the novels and Wikidata provides coordinates to many locations, we can write a federated query to retrieve those and visualize the locations on a map.

First you need to define the prefixes you are going to use for the other knowledge base in the query as:

PREFIX wd: <http://www.wikidata.org/entity/> #wikidata wd PREFIX wdt: <http://www.wikidata.org/prop/direct/> #wikidata wdt

Within the WHERE-part you can query all items you want to as long as they have a P13 (exact match with Wikidata) property. So in our example ?item mmdt:P32 ?nar_loc. #a novel has a narrative location ?nar_loc mmdt:P13 ?WikiLink. #the narrative location has an exact match.

Next using the SERVICE referring to the Wikidata SPARQL endpoint, we can get all information listed on the matching Wikidata entry. To write the triple, we now need to use the property values of Wikidata. Here P625 is the coordinate location of the Wikidata entity.

SERVICE <https://query.wikidata.org/sparql> { ?wikidataEntityLink widt:P625 ?coordinateLocation. }

Example: Show all narrative places (using the coordinate locations property of Wikidata)

Federated query
: SPARQL can be used to express queries across diverse data sources, whether the data is stored natively as RDF or viewed as RDF via middleware.

Previous Next

Where to find…
▼

… MiMoTextBase: data.mimotext.uni-trier.de
… SPARQL-Endpoint: query.mimotext.uni-trier.de
… MiMoText project site: https://mimotext.uni-trier.de

Am I right here or Where do I start?
▼

If you are new to SPARQL, you can go through the (short)Tutorial,which will give you an overview of how to write basic queries based on examples inMiMoTextBase. It’s supposed to give newbies an introduction to SPARQL, but it cannot give you a deep knowledge of SPARQL – maybe theseresourcescan help you with that.

If you are interested in MiMoTextBase and its content onauthors,novels,spacesorthemesof the French novel in 1751-1800 with already some SPARQL knowledge, you can have a look at the links.

WithinGOING FURTHER there are some queries on the data containing overviews of items like dates of publication or themes changing over time and comparing the different sources of the data inMiMoTextBase together with some interpretation on the outcome which could show the potential of initial questions on further research.

If you want more detailed information about the structure and the aims of our tutorial, you can find it in theintroduction of the tutorial.Information on the infrastructure and the models behind MiMoTextBase you can findhere.

I have no results. What can I do?
▼

Having no results in the result table can have different reasons. A simple solution is to check whether the variables are spelled the same in the SELECT and the WHERE part of the query.

Another reason could be being too specific in the query. Not all items in MiMoTextBase contain all information on all properties due to its sources. So it can be helpful to add the OPTIONAL function on some of the properties in your query, seehere.

Error message: Bad aggregate
▼

If you run into this error message, you probably have to group items. In the example below, we use the count function, but forgot to add GROUP BY.

Query to retrieve count of published works per author:

The solution is easy: We have to aggregate ?authorName by grouping. We can now get the results in descending order via order by desc(?count) and set a limit of 20 to get the top 20.

Query to retrieve authors with most novels published (top 20):

I have too many results. What can I do?
▼

Sometimes you can get many results on a query which can slow down the result generation or impair the readability of some visualizations. In those cases you could add the LIMIT-operation (seehere)to only get the TOP x items or the HAVING COUNT-operation (seehere)if you want only results that lie above a certain threshold.

If some of the items appear more often in the results than they should, make sure you filter all labels for one language (FR, EN, DE) separately as the graph is multilingual and the output will represent all languages within the graph, seehere.

How to find the right item / right property?
▼

If you're looking for the right identifier for properties, novels, authors, themes or locations, the simplest way is to visitdata.mimotext.uni-trier.deand type in the label (for example “London” or “about” or “philosophy”) in the search bar. The numerical identifier of the property or the item is visible in the URL or behind the name of the item or the property.

You can also consult our lists of themes, locations and properties and their numerical identifier in the knowledge graph below.

List of properties

Query:Retrieve a list of all the properties used in this graph

List of themes

For a list of all thematic concepts in the graph, see thisquerywhich lists all thematic concepts and their Q-identifier, ordered by occurrence:

List of locations

For a list of all narrative places in the graph, see thisquerywhich lists all narrative places and their Q-Identifier, ordered by occurrence:

These queries list themes or locations ordered by occurrence. We recommend using items or properties which have a certain number of connections in the graph, in order to get good results (with enough data points).

The query is very slow or there is a time out. What can I do?
▼

There are several possible reasons for a slowdown or a timeout of your query. It could be that the quantity of results is very high, so you might limit the results to check if the syntax of the query is OK. This is done by using theLIMITparameter. The LIMIT tells the algorithm where to stop, so if you insert for example LIMIT 100 at the end of your query, it will stop after 100 results. This can be helpful for debugging.

Parameters which potentially slow down the query are DISTINCT or ORDER BY. A strategy might be to comment them out to see if these slow down your query.