AMIE 3.5.1

Fixed issue with RDF file formats being rejected

AMIE 3.5

Use with remote knowledge base server

Since loading and storing knowledge graphs can take a significant amount of memory space and time, AMIE 3.5 makes it possible to run the mining routine against a remote knowledge base, splitting the architecture into two parts communicating over network.

Below is a basic setup example to use AMIE with a remote knowledge base.

Server-side

java -jar amie3.5.jar -server [TSVFile] -port <Server Port (default: 9092)>

This will load the data into the memory of the server.

Client-side

java -jar amie3.5.jar -client -serverAddress <Server Address (default: localhost:9092)>

In this case the client will mine the rules on the server deployed at the provided answer.

NOTE:

• Client and Server communicate using the WebSocket protocol.

Optional: Enabling cache

AMIE may run the same query more than once. It is therefore possible to enable query caching for either server or client side with the -cache option. This option is available only for remote mining. The cache option can be set either on the client or on the server side. The cache is automatically saved upon shutdown. If a corresponding cache is found, cache save is loaded, unless -invalidateCache is passed as argument.

The cache can improve performance significantly by reducing the amount of queries sent over network or executed by the KB.
Performances will vary depending on the knowledge graph and the user parameters.

The performance of the cache and the remote setting is sensitive to the data, as this defines the size of AMIE's search space as well as the amount of queries and query answers that will be sent over the network.

NOTE:

• Cache uses Least Recently Used (LRU) policy. As of yet, only LRU cache policy has been implemented.
• Custom cache policies can be implemented in amie/data/remote/cachepolicies package.
• Cache is saved locally in the cache directory using the knowledge graph file name and run options.

AMIE 3

AMIE 3

This release contains AMIE 3, using an integer in-memory database.

It contains the optimizations:

• Laziness: /!\ Not on by default, use "-bias lazy".
• Variable order: head variable by default, see "-vo" option.
• Existential Detection.

AMIE 3 - ByteString KB

AMIE 3

This release contains AMIE 3, before migration to an integer in-memory database using fastutils.

It contains the optimizations:

• Laziness: /!\ Not on by default, use "-bias lazy".
• Variable order: head variable by default, see "-vo" option.
• Existential Detection.

AMIE+

AMIE+

Migrated to a Maven project.
Continuation of project amie-old. Please refer to this repository for code history.

The bugfix of the printing thread is in a later release. Please use option "-oute" to avoid potential infinite loop.