{"id":1025,"date":"2017-07-20T09:53:33","date_gmt":"2017-07-20T09:53:33","guid":{"rendered":"http:\/\/www.kelp-ml.org\/?page_id=1025"},"modified":"2017-08-31T16:25:16","modified_gmt":"2017-08-31T16:25:16","slug":"generating-input-data","status":"publish","type":"page","link":"http:\/\/www.kelp-ml.org\/?page_id=1025","title":{"rendered":"Generating Input Data"},"content":{"rendered":"

To generate the input data for KeLP we developed a specific project: kelp-input-generator<\/a>.<\/p>\n

This project relies on third party software components, such as the Stanford Parser<\/a>, and provides the functionalities to extract KeLP data structures from text snippets. Being a general purpose machine learning platform, KeLP is not limited to only Natural Language Processing tasks. However, for the moment, we do not provide any feature extraction capability for different fields.<\/p>\n

In order to preserve the lightweight of the main KeLP project,\u00a0kelp-input-generator<\/a>\u00a0is not included in kelp-full<\/a>. If you want to use the kelp-input-generator<\/a> functionalities in your maven project you can easily include it with the following Maven repository:<\/p>\n

\t\t\r\n\r\n\t\r\n\t\t\r\n\t\t\tkelp_repo_snap<\/id>\r\n\t\t\tKeLP Snapshots Repository<\/name>\r\n\t\t\t\r\n\t\t\t\tfalse<\/enabled>\r\n\t\t\t\talways<\/updatePolicy>\r\n\t\t\t\twarn<\/checksumPolicy>\r\n\t\t\t<\/releases>\r\n\t\t\t\r\n\t\t\t\ttrue<\/enabled>\r\n\t\t\t\talways<\/updatePolicy>\r\n\t\t\t\tfail<\/checksumPolicy>\r\n\t\t\t<\/snapshots>\r\n\t\t\thttp:\/\/sag.art.uniroma2.it:8081\/artifactory\/kelp-snapshot\/<\/url>\r\n\t\t<\/repository>\r\n\t\t\r\n\t\t\tkelp_repo_release<\/id>\r\n\t\t\tKeLP Stable Repository<\/name>\r\n\t\t\t\r\n\t\t\t\ttrue<\/enabled>\r\n\t\t\t\talways<\/updatePolicy>\r\n\t\t\t\twarn<\/checksumPolicy>\r\n\t\t\t<\/releases>\r\n\t\t\t\r\n\t\t\t\tfalse<\/enabled>\r\n\t\t\t\talways<\/updatePolicy>\r\n\t\t\t\tfail<\/checksumPolicy>\r\n\t\t\t<\/snapshots>\r\n\t\t\thttp:\/\/sag.art.uniroma2.it:8081\/artifactory\/kelp-release\/<\/url>\r\n\t\t<\/repository>\r\n<\/repositories>\t\t\t\r\n\t\t\r\n\t\t\r\n<\/pre>\n
Then, the Maven<\/a> dependency for the kelp-input-generator project is:<\/p>\n
\t\t\r\n\r\n\r\n\t\t\r\n\t\t\tit.uniroma2.sag.kelp<\/groupId>\r\n\t\t\tkelp-input-generator<\/artifactId>\r\n\t\t\t1.0.1-SNAPSHOT<\/version>\r\n\t\t<\/dependency>\r\n<\/dependencies>\r\n\r\n<\/pre>\n
Currently, kelp-input-generator<\/a> allows to easily generate TreeRepresentation<\/a>s from text snippets. In particular, it provides the capabilities to extract the LOCT, LCT and GRCT representations, which are a tree views of a dependency graph, as introduced in (Croce et al., 2011).<\/p>\n
KeLP uses its own format for representing graph data. However, a converter from the popular gSpan format is available inkelp-input-generator<\/a>: it.uniroma2.sag.kelp.input.graph.GspanFormatConverter. The main method on the class can be invoked passing as parameter the gSpan file (and optionally a file with the target labels if they are available and they are not included in the gSpan file).<\/div>\n
If your input graphs are in a format supported by Open Babel<\/a>, the following script<\/a>\u00a0converts graphs from one of the Open Babel to gSpan. Therefore, all 111 Open Babel formats are indirectly supported as well.<\/div>\n
In the future we plan to extend kelp-input-generator<\/a> by adding the possibility to extract shallow and constituency tree representations, as well as SequenceRepresentation<\/a>s and DirectedGraphRepresentation<\/a>s.<\/p>\n
References<\/h3>\n
Danilo Croce, Alessandro Moschitti, and Roberto Basili.\u00a0Structured lexical similarity via convolution kernels on dependency trees<\/em>. In Proceedings of EMNLP, Edinburgh, Scotland, UK., 2011.<\/p>\n","protected":false},"excerpt":{"rendered":"
To generate the input data for KeLP we developed a specific project: kelp-input-generator. This project relies on third party software components, such as the Stanford Parser, and provides the functionalities to extract KeLP data structures from text snippets. Being a general purpose machine learning platform, KeLP is not limited to only Natural Language Processing tasks. Read more about Generating Input Data<\/span>[…]<\/a><\/p>\n","protected":false},"author":1,"featured_media":0,"parent":0,"menu_order":0,"comment_status":"closed","ping_status":"closed","template":"","meta":[],"_links":{"self":[{"href":"http:\/\/www.kelp-ml.org\/index.php?rest_route=\/wp\/v2\/pages\/1025"}],"collection":[{"href":"http:\/\/www.kelp-ml.org\/index.php?rest_route=\/wp\/v2\/pages"}],"about":[{"href":"http:\/\/www.kelp-ml.org\/index.php?rest_route=\/wp\/v2\/types\/page"}],"author":[{"embeddable":true,"href":"http:\/\/www.kelp-ml.org\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"http:\/\/www.kelp-ml.org\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=1025"}],"version-history":[{"count":14,"href":"http:\/\/www.kelp-ml.org\/index.php?rest_route=\/wp\/v2\/pages\/1025\/revisions"}],"predecessor-version":[{"id":1062,"href":"http:\/\/www.kelp-ml.org\/index.php?rest_route=\/wp\/v2\/pages\/1025\/revisions\/1062"}],"wp:attachment":[{"href":"http:\/\/www.kelp-ml.org\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=1025"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}