To achieve an e ective and e cient way of disseminating information to ever growing communication channels, we propose an approach that separates the information and communication channels and interlinks them with an intermediary component. The separation enables various dimensions to reuse the information and communication channels in transactional communication. In this paper we introduce our online communication platform, which is comprised of several components. The important roles of semantic web technologies to the platform are explained in detail, including a use case to show the contributions of semantic web in supporting the e ectiveness and e ciency of information dissemination.
In today's internet era, the number and kinds of information dissemination channels are growing exponentially and changing constantly. Websites, e-mails, blogs and social media have become the mainstream means of communication. Nevertheless, information dissemination is not only about nding suitable channels, but also tting the content to the available channels. These are the main challenges for e ective and e cient information dissemination, and for online communication in general.
Our solution to overcoming these challenges is to decouple information from
channels, de ning separate models for each of them, and then interlinking them
with an intermediary component [
In this paper, we focus on the information modeling (including annotations) part such that the matchmaking of information to appropriate channels can be performed e ciently. First, we present the overall architecture, then we discuss how semantics contribute to the solution and nally we show a use case, followed by the conclusion and future works.
Shown in Fig. 1, the online communication platform consists of several components which are grouped based on their conceptual functions:
Information Management is responsible for gathering the content from data sources (annotated and un-annotated) and representing them into the common vocabularies. First, the contents are extracted by a Content Extractor (implemented using Any23 1), then stored onto a Triplestore such as OWLIM 2. Further, an RDF to OO Mapper (implemented using RDFBeans 3) maps the stored triples onto object-oriented models to be used by the other components. For annotated sources where the sources have been annotated with the selected vocabularies, the content can be extracted automatically. For un-annotated sources, a manual mapping is required to inter-relate the database items (i.e. table elds) to relevant terms in the desired vocabularies.
Weaver is responsible for matching the information to appropriate channels through a rule based system. A Rule Editor enables experts to create and maintain rules through an integrated user interface and access-controlled rules repository. The rules are then matched to the facts in the working memory of the rule-based system by a Rule Engine. In our implementation we use Drools 4.
Channel Management is responsible for distributing the information to
the selected channels according to the de ned rules. Dacodi 5 o ers various
functionalities for distributing the content to the selected communication channels,
as well as for collecting and analyzing feedback from those channels [
2http://www.ontotext.com/owlim 3http://rdfbeans.sourceforge.net
3
Semantic technologies contribute mainly to content modeling, namely in how to obtain content from distributed and heterogeneous sources (i.e. through annotation) and represent them in a common representation to make an e cient match between information and desired channels possible. The matching is not between content sources but between the common representation to channels.
To achieve a reusable and interoperable information model, we selected vocabularies (whole or partial) from the Linked Open Vocabularies 6: 1. Dublin Core 7, all metadata terms to support resource description 2. Friend of a Friend 8, a vocabulary to describe people, the links between them, the things they create and do 3. Good Relations 9, a vocabulary to describe e-commerce products and services 4. Schema.org 10, a collection of tags to markup a page in ways recognized by major search engines These vocabularies are widely used, especially Schema.org which has been adopted by webmasters to increase their webpages' visibility in search engines.
We show these contributions in detail within the Tourismusverband (TVb)
Innsbruck 11 use case. As one of the big tourism boards in Austria, its goal is to
achieve the highest visibility possible in search engines as well as to be present in
various social channels [
14.07.2014 - 01.01.2015</div> <time itemprop="endDate" datetime="2015-01-01T12:00"/> <span itemprop="location" itemscope itemtype="http://schema.org/PostalAddress">Location: <span itemprop="streetAddress" content="Markthalle">
Markthalle (Herzog-Siegmund-Ufer 1-3, AT-6020, Innsbruck)</span> <meta itemprop="streetAddress" content="Herzog-Siegmund-Ufer 1-3"/> <meta itemprop="addressRegion" content="Innsbruck"/> <meta itemprop="postalCode" content="6020"/> <meta itemprop="addressCountry" content="AT"/></span> </div> In this example, information about Event is annotated with the term Event from Schema.org by using microdata format 12. 6http://lov.okfn.org 7http://dublincore.org 8http://www.foaf-project.org 9http://purl.org/goodrelations/ 10http://schema.org 11http://www.innsbruck.info 12http://www.w3.org/TR/microdata/ b) The publication rules were de ned to guide the publication of extracted contents to selected channels. rule "Event Publication Rule" when item : Event() then insert(new ItemToBePublishedIn(item, facebookWall))
insert(new ItemToBePublishedIn(item, youtube)) end In this rule, each time a new Event was found in the extracted contents, it was then prepared to be published to the facebookWall, youtube (instances of TVb's Facebook and YouTube accounts respectively). 4
In order to evaluate our work, we compared the number of visitors to the TVb's website before and after annotating the content. Compared to the same period in 2013, the number of visitors increased by 8.63% between Jan-Feb 2014, which may be caused by the annotation. Also, the platform is currently being tested by 6 people at TVb as a substitution to their social media dissemination tool.
The platform was comprised of several components and used semantic web technologies to integrate various information sources, extracting and representing the content into common vocabularies to enable e cient matchmaking to appropriate channels using a rules-based approach. There are four vocabularies currently supported and in the future, we would like to add more vocabularies (i.e. Schema.org Action, SIOC 13) to enhance the channel management, in order to improve the feedback collection, for example.
Acknowledgements We would like to thank all members of the OC working group 14 for their valuable feedback. This work was partly funded by the EU FP7 under grants no. 600663 (Prelida), 257641 (PlanetData) and 284860 (MSEE).