← Go back

LIMES

Link Discovery Framework for Metric Spaces

About the demo

Using the Web UI

LIMES Web UI is an additional tool to write configuration file in the XML using web interface and execute it using the LIMES server. LIMES Web UI consists of six main components: prefixes, data source / target , manual metric, machine learning, acceptance/review conditions and output.

1 - Prefixes

The Prefixes component consists of two parts:

  • Currently added prefixes. They look like chips, containing the label of the prefix and a hover tooltip with the namespace.
  • Add new prefix (optional). In most cases, our interface is able to automatically find the common prefixes from prefix.cc. In case the user want to add a prefix manually, (s)he can type the prefix, choose the its respective URI from the dropdown list (if any) or manually type it. Then, the user click on Add and the prefix will be added as a new chip. Again, this process is necessary for common prefixes.

2 - Data source / target

The Data source and data target consists of the two similar components, which include three input fields:

  • Sparql endpoint/Local file: One of two options can be chosen. Sparql endpoint means that the user will select the endpont from the list. Local file means that the file should be provided as an endpoint.
  • Endpoint: A dropdown list of available endpoints. Moreover, the user can try to search for the endpoint, typing it in the input field or write your own endpoint. After clicking on the endpoint from the list or writing it by hand and press the Enter, the user will get the list of restriction classes according to this endpoint.
  • Restriction: Contains of three parameters splitted by space (?s rdf:type some:Type). The third parameter will be changed automatically after changing the restriction class.
  • Restriction class: A dropdown list of restriction classes according to the endpoint. You can start typing the name of the class and the list will be filtered automatically. After choosing the restriction class, you will get all the properties related to this class.

3 - Manual metric and machine learning

After the data source and target have chosen and you have got a message that properties have already received, you are now ready to build a metric. You can build a manual metric or use machine learning. These options are interchangeable. Consider first the manual metric tab.

Manual metric

To build a metric you can drag and drop elements from the toolbox.

There are 8 blocks:

  • Source property,
  • Target property,
  • Optional source property,
  • Optional target property,
  • Measure,
  • Operator,
  • Preprocessing function,
  • Preprocessing rename function,
  • Complex reprocessing function.

In the Workspace you can see the Start block, which cannot be removed. Building the metric exactly starts from connecting the Measure or Operator block to the Start block. Once you connect the Measure block, next step can be connecting the Source property and the Target property. The Measure block consists of list of measures and checkbox for enable changing the Threshold value.

Optionally, you could preprocess your Source and Target properties using the Preprocessing function. For instance, to use Preprocessing rename function you should put the Source property or Target property into this block and then you can connect it to the Measure. Using Complex reprocessing function, for example "Concat", it is possitble to concatenate the values of two properties and rename the final value.

An Operator block accepts two Measure blocks as inputs. i.e., two Measure blocks must be attached to its input ports.
The Operator block includes the list of operators.

Optional source property and Optional target property can be connected just after connecting the Source property and the Target property.

In general, once you add Source property/Target property/Optional source property/Optional target property to the workspace, you can choose the property from the dropdown list. Prefixes of the properties will automatically added to the Prefixes. There is a checkbox PP  for each property, which defined as Property path. If user checks this PP, it means that the mode Property path is enabled and new subproperties will be loaded after some time if they exist. The user can uncheck PP   and then the property will return to the previous state.

You can remove the block using the trash can or using right click on the block.
On the bottom after the Workspace you can see the formed Metric.

At the top before the Workspace there are two options related to Workspace:

  • Export workspace to xml: You can download an xml file of the current Workspace with connected blocks.
  • Select file for the importing to the workspace: You can upload the xml file of the saved Workspace to change the current Workspace.

Machine learning

The tab Machine learning consists of three parts:

  • Name:The machine learning (ML) algorithm name. The currently available ML algorithms include WOMBAT Simple, WOMBAT Complete and EAGLE
  • Type: The ML algorithm types, which include supervised batch, supervised active and unsupervised
  • Parameters: A list of parameters which can be added to the currently selected ML algorithm.

If Type is supervised batch or supervised active, you will see the additional input, where you can upload the file with training data.

4 - Acceptance and review conditions

In this component you can define the Acceptance Threshold and the Review Threshold. In addition, you can rename the names of files, which can be created after execution. Besides, you can change the Relation. Instead of prefix you can write the namespace and its respective URI will be automatically found by the interface, converted to a prefix (if it exists in prefix.cc, otherwise you have to manually add it).

5 - Output

Here you can choose an output format, including turtle (TTL), n-triples (N3), tab separated values (TAB), comma separated values (CSV).

Display config and run

There are three buttons at the bottom of the page: Display config, Execute, Check the state of the previous run.

  • *If you click on the Display config, you can look at formed xml config. Also, if you want you can save it.
  • *The Execute button will immediately start the process of executing this xml config. You can have a look at the log messages by clicking on the link Show log.
  • Check the state of the previous run button. The execution can take time. To look at the result later, you must copy the execution key. In order to get the result of the previous run, you should click on Check the state of the previous run button. Here you can paste the execution key, which you copied when you ran the execution. Then click on Check and you will get the result. If the run is finished without errors and result is not empty, then you can download accepted links and reviewed links files.