Link Discovery Framework for Metric Spaces
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.
The Prefixes component consists of two parts:
The Data source and data target consists of the two similar components, which include three input fields:
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.
To build a metric you can drag and drop elements from the toolbox.
There are 8 blocks:
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:
The tab Machine learning consists of three parts:
If Type is supervised batch or supervised active, you will see the additional input, where you can upload the file with training data.
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).
Here you can choose an output format, including turtle (TTL), n-triples (N3), tab separated values (TAB), comma separated values (CSV).
There are three buttons at the bottom of the page: Display config, Execute, Check the state of the previous run.