Scientific and Technological AimsSEMAMO aims at building a technological framework that allows the automatic capturing of market information and the semi-automatic analysis of these data for market research and benchmarking. To automate this process, online data harvesting, semantic domain modeling, and semantically guided (i.e., domain specific) analysis and reporting technologies are integrated. This framework will be tested in the area of tourism, but can be easily transfered to other sectors due to the underlying domain model. More specifically, SEMAMO aims at: - building a metadata model that captures the required semantics to enable the interpretation of the market data according to application needs;
- sampling of data from Web sources: this will include the learning of properties of the data sources (such as data change frequency) in order to guide the samplig process as well as to update the domain model, making the entire system adaptive;
- developing an ontology-driven strategy for cleansing of data; and
- developing semantic annotation and analysis methods for product as well as market data.
On a methodological level, SEMAMO will use a semantic approach to adapt practices of market research to the specific requirements of online markets. This includes the important question of how to gather relevant empirical market data from online sources. In this respect, two main problem dimensions have to be addressed, viz. - the assignment of “harvested” product (and attached price) information to identified products, or markets in the narrow sense, for the purpose of direct price comparison (formally, i.e., a classification task assigning collected data to product classes based on product descriptions wrapped from the Web sources), and
- the (adaptive) decision on how often individual online sources (expressed as URLs) have to be sampled in order to optimise the data collection effort (i.e., to keep the price forecast intervals balanced) by adaptive sampling schemes.
Starting with the assumption that the Web has grown into a faithful mirror of the (economic) world, and that, hence, decision-relevant market information can indeed be gathered from the Web and analysed automatically, SEMAMO is an interdisciplinary research effort integrating semantic web technologies and statistical methodologies: the customization and automation of the empirical analysis process will be facilitated by explicit application domain models expressed in a suitable formalism comprising the respective subject-matter (domain) ontologies. This engenders not only a plain combination, but rather fresh research into the effective integration of both parts. In order to obtain reasonable flexibility of the approach, SEMAMO distinguishes clearly between generic function components independent of specific market analysis tasks, and a configuration component stating the “semantics” of individual analysis domains (such as, e.g., travel package offers). In brief, the specific technological and scientific aims tackled in the project are as follows: - Developing an ontology-driven strategy for data cleansing and rectification assuring that only comparable products and market offers are pooled for comparative analysis.
- Learning of sampling strategies of data from Web sources: SEMAMO addresses the need to determine sampling strategies adaptively to generate results with a required confidence value.
- Learning of empirical properties of the data sources (e.g., data change frequencies) as well as new structural characteristics (e.g., relationships, new product [descriptions]). This will lead to an update of the semantic domain model, making the system adaptive to changing markets.
- Devising a data aggregation model according to interesting aggregation hierarchies, i.e. geographic regions, product categories, bundles, etc., given a specific domain model.
- Developing semantic annotation and analysis of product as well as market data, e.g., distribution of offers over a certain time period or over geographic regions.
- Developing and implementing semi-automatic analyses – again based on semantically controlled application of statistical analysis functions – of the harvested data.
- Building a metadata model that captures the required semantics to enable the interpretation or analysis of the market data according to application needs. The focus will be on a generic framework that can be applied to a number of domains.
- Identification of business critical decisions and how the decision finding process can be supported by analysis and reporting tools.
In all this, the essential core of SEMAMO consists in automating the market data analysis process flow starting from harvesting online sources through data rectification up to analysis such that both, the data harvesting process and the data analysis/reporting stages are directly driven by the subject-matter relations defining markets, products, and relevant decision criteria encoded in the semantic domain model of an application. In particular, SEMAMO embodies an adaptive approach to empirical Web data analysis by feeding back derived information (product occurrences, description elements, price variation) into the harvesting and reporting processes, respectively, based on constantly updated domain representations. In order to achieve its ambitious goals, SEMAMO is based on a range of fields such as Web data extraction, data cleansing and record linkage, (statistical) data analysis and reporting, as well as semantic domain modelling, and market research. The objective is to develop a self-contained framework that allows to better understand, analyse, and report on developments in e-markets. For turning the research effort of SEMAMO, in the longer run, into a marketable tool matching the needs of a wide variety of application domains, e.g., travel industry, electronic consumer goods, fast moving consumer goods, etc., a flexible and scalable component architecture is envisaged. Eventually, as many industry sectors and application scenarios as possible should be covered. Within the SEMAMO project, the domain of e-tourism will be used as a test case, based on the rationale that this market area definitely poses a non-trivial challenge due to its complexity. Economic Relevance and PotentialIn the Internet-based economy, traditional market research, including methods of market segmentation and price discrimination, no longer work the way they used to. In particular, the transparency of e-markets and the speed of market changes call for an increasingly comprehensive and quicker monitoring of markets and competition. A natural response to this overall development, advanced information technology – and particularly semantic technologies – provide a means to expand the range of and to accelerate market observation (on both global and local business levels) by reducing the cost of information procurement and, thus, speeding up competitive decision-making. In this respect, SEMAMO arguably extends the scope of current business intelligence methodologies and solutions, exploiting the vast potential of semantic technologies to solve a problem of tremendous practical economic relevance untackled as yet. Expectedly, the SEMAMO partners will be in an excellent position to leverage project outcomes both scientifically and economically. The economic advantages for potential customers will clearly result from the improved visibility of the electronic distribution channels and the online markets as a whole. According to a 2006 Gartner Research report, on average, a 1% improvement in price translated to an 11% increase in profitability. By contrast, according to the same report, a 1% improvement in fixed costs or in variable costs only increases profitability by 3% and 7%, respectively. SEMAMO will gather market data required to calculate optimal prices, often leading to substantial price improvements. The sheer extent of the online market channels does no longer allow for a manual observation and analytical activities based on cut-and-paste. In contrast, a systematic approach is required to deliver reliable and predictable results that can be used on a daily bases. Complete market coverage is required to meet the needs of involved product managers/pricing experts. SEMAMO will provide this coverage by a systematic approach towards market intelligence applications. Moreover, the near real time quality of SEMAMO responds to the fast moving online markets such as travel, perishable goods, and consumer goods. Customers of the envisioned solution will be able to react faster and with improved accuracy to market movements, and thus put in the forefront of a more competitive market appearance. The target market for online market monitoring is defined as all companies who intend to or already play significant roles in the relevant online and also offline markets. A further important criteria to segment those companies is their current skills in online marketing and their willingness or market pressure to further develop this ability. Specifically large online vendors will be the most likely targets for the potential product. From experience with related systems we can already forecast that usually the companies with the largest number of business processes to handle and a high degree of organization are usually the first to understand the benefits of SEMAMO. Industry sectors that will pick up such a solution are characterized by their affinity to online market channels. As of today, the travel industry shows the highest degree of online business. Other industries are electronic consumer products, computer parts, cars and car parts, furniture, fashion, etc. The US E-Commerce retail market values for USD 220 billion in total (Forrester Research Inc., 2007). Regionally, among the segments targeted the German speaking market ranks first, with the French and UK market following next. The US market is of course very interesting due to its e-commerce maturity, however significant investment is needed too. Partner
|
Walchhofer, N., EC3, Pöttler, M., Werthner, H., Vienna University of Technology, Semantic Market Monitoring in Tourism, Journal of Information Technology & Tourism Workshop Series, October 2008, Vienna. (download article) Walchhofer, N., Fröschl, K. A., Hronský, M., & Hornik, K. (2009a) Dynamic population segmentation in online market monitoring, Proc. IFCS 2009 (Dresden, Germany), to appear. (download article) Walchhofer, N., Hronský, M., & Fröschl, K.A.(2009b) The online market observatory: A domain model approach. In: Karagiannis, D. & Jin, Z. (eds.) Proc. KSEM 2009, 229-240. (download article)
|
Deutsch
English
Profil
