Technology-oriented companies are typically interested in monitoring developments concerning their technologies. However, most companies, especially SMEs, don't have an e cient process how this is achieved. If at all, e orts are mostly limited to uncoordinated keyword queries on web resources. Here, we present a semi-automatic approach that allows for structured and continuous detection of relevant, novel and domain speci c documents appearing on the Web. Our system is based on a semantic wiki where the domain expert is able (i) to store all relevant information in an adequate knowledge base with the ability for monitoring and trend mining and (ii) to import detected novel items such as future technologies and their properties to the knowledge base in a continuos fashion. The latter is achieved by generating a structured query based on the user context and by representing found documents as semantic graphs. In this way, novel items can be found easier and in a semi-automatic fashion.
Technology forecast and trend detection are indispendable tasks for technology companies in order to be informed about market developments and inventions in their elds. With the advent of more and more documents on the Web, This work is supported by the German Federal Ministry of Education and Research (BMBF) under grant 02PJ1002 (SyncTech). companies face the task of extracting relevant and novel information for this purpose. Currently, this has to be done usually purely manually and without any structured background data making it a very time-consuming task. Therefore, we provide a semi-automatic process for trend detection and monitoring services. We present a semantic wiki-based application which is based on ontology-based information extraction (OBIE) where ontologies are used within the information extraction (IE) process. Since usually appropriate ontologies regarding technologies and their properties are missing or are too small, we focus our work on the crucial task of how to e ciently nd new textual information which is relevant to the domain expert, but has not been stored in the knowledge base (KB) and, therefore, has been made usable in some sense.
Within the TREC \novelty track" in 2002{2004 [
Figure 1 gives an overview of the interplay between an ontology and documents with potentially novel information: Given our own KB with instances and schema, our goal is to search for documents and to rank them, so that the documents most novel to the KB and relevant to both the query and the KB have the highest ranking. In a second step the user is able to import phrases marked in the document into his/her KB as property values.
Concerning the rst part, Semantic MediaWiki1 as an instance of a semantic wiki is assigned the central role: The user is able to create new wiki pages (within the semi-automatic process or just manually) and to add Lithium-ion battery
Priciple Assessment Sources/Contact WWW search Technology description ontology population GUI-2 fact extraction
GUI-1 document ranking auto annotation appropriate properties with the help of a class-speci c form (see gure 2). Internally, all data is stored in a structured way. The wiki allows the user to create a search query out of the context by taking instances and property values (from the KB) as well as search keywords written by the user. After an optional expanding of the query graph with neighbouring entities, we can generate the nal query graph. Since all documents are annotated with the help of named entity recognition tools2, we can compare the generated query graph with all document entity graphs (generated from extracted named entities). Ranking of the documents is facilitated by weights which were assigned to every relation in the KB schema graph. We can use implicit user feedback in the following way: If a user imports some novel item as a new property or instance, the weights in the KB schema graph are adapted. By this means, we can defer to the personal views what relationships between certain classes and properties (or other classes) are of great signi cance and should be reinforced for next search sessions. Our focused use cases are determined by our use case partners3 which are medium-sized technology companies. Hence, the lightweight ontologies we used consist of classes like technology, institution, and product. As document corpus, web documents retrieved by search engine requests are considered. In addition, trend detection in conjunction with patents can be enabled by using the patent database Espacenet4, where access to over 70 million patent documents and their meta data is provided.
Existing processes and tools for trend mining and technology watch are often only rudimentary implemented, especially in SMEs. We have presented a semantic wiki for storing and displaying structured information about a speci c 2One of these tools is the wikify service of the Wikipedia Miner (http://wikipedia-miner.cms.waikato.ac.nz/) which we adapt by using the content of our domain speci c semantic-based wiki. In order to detect also new entities, property values, and relationships, we use GATE (http://gate.ac.uk), a well-established rule-based framework. 3Industry partners within the German research project syncTech (http://synctech-innovation.de). 4http://worldwide.espacenet.com PrincipleAssessment Sources/Contact
Technology description:
A Lithium-ion battery (also: Li-ion battery) is a hypernym of batteries on the basis of lithium.
Energy Storage Independency of time and place. Very high energy density. Thermal stable. No memory effect.
Industry, Household, Automotive, Other easy A Lithium-ion battery (also: Li-ion battery) is a hypernym of batteries on the basis of lithium.
Energy Information
Transportation Storage (b) Figure 2: Screenshots of a Semantic MediaWiki: (a) displaying technology property values within a wiki page (b) edit functionality using form. domain (industrial technology eld) and for generating a context-aware semantic search query. With the help of a new proposed ranking schema, the more relevant and potentially novel information a document contains, the higher it is ranked and, hence, more likely to be worth reading and used for ontology population. Due to the use of structured information and approriate background data the way of doing trend mining can be changed towards a semi-automatic process with better search and monitoring capabilities.