their french competitor Grandcouleur S.A. tion of Grandcouleur S.A. and immediately, you meet ible delivery conditions. Nice strategy! basic material industry, GrandBasic Inc., your biggest Consider the case you are key account manager of Colvelopment, you discuss with your production manager the German Bigcolor AG is in acquisition talks with orBasic Inc, a supplier of basic material for the color action and your fast and exible supply conditions, convince your key account at Bigcolor AG to enhance So what? In a rst happy reaction about this deduction manager conrms a possible increase of production and prepares a quick switch of production line. your key account at Bigcolor AG to get his agreement your supply agreement also for their new subsidiary industry, responsible for several key accounts in the immediately after aquisition. Nice analysis! Your proTwo month later, you read the news about the acquisinewspaper article, that one of your biggest customers, European color Industry. The other day you read in a for higher sales volume of basic material with very exof basic material in a short time period. You plan to but tells you, that Bigcolor AG acquired Grandcouleur Your key account is impressed about your quick rethe ability of your company to increase the production S.A. due to the fact that they have a subsidiary in the By this way one may introduce an abundance of Information archives are document-based. For a count for only few of many aspects that occur in a particular document (e.g. only the information pieces of information again. redundancies on the one hand and one may accollective gathering of facts this view is too coarse. given in the headlines), thus missing the crucial The next problem stems from the fact that typically document management systems rely almost exclusively on information retrieval techniques, which are rather unreliable. if background knowledge is used, but document These implications may only be made transparent documents, it is most often too cumbersome to tion is that this unit may no longer be the sole one company sells one of its units, the implicadescribe the relevant implications, too. E.g., if Even if convential annotation is used for these supplier for the overall company as it was before. players permanently restructure their organization, try News! Newspaper articles, online databases and annual comNow, it is easy to recognize, that your analysis of competitor. Therefore, Bigcolor AG made a vertical the management, published in reports and distributed.
Nowadays, the market analysts, consultants and inpany reports were analysed and competitors web pages to acquire small innovative technology leaders or gh t is no day without news about mergers, acquisitions Grandcouleur S.A. and GrandBasic Inc. Even worse, unfortunately, instead of higher sales volume you wont integration and bought very interesting supply conactivities of their industry with traditional methods. industry. One of the most dynamic markets in the past There exist several problems in this research process: but how should you handle such situations? There is industry. egy. Later on, you will hear a lot of colleagues telling ditions via the acquistion of Grandcouleur S.A. Thus, mation about the acquisition to create a better stratthe situation was wrong. You had not enough inforwere thoroughly tracked. The results are presented to and a lot of other strategic moves of players in every namic situation leads to the problem, that no market participant has a permanent clear picture about the you stories about the well-known relationship between house market research departments try to track the years has been the chemical industry. Here, big global about entry strategies in new markets. This very dyan ever faster pace of change in global markets. There have any sales in near future with Bigcolor AG. Bad business case that the history analyzer may be used for a nal step we describe how the overall system has it oers on the querying and on the pro visioning side. ontology with an elaborate temporal structure. that are interesting in themselves and which are comCHAR constitutes an new type of system that is inIn the rest of this paper we will rst elaborate the range of ontology-based knowledge management tools knowledge base that support corporate researchers and nical point of view. Seen from the side of the users, been conceived and implemented (Section 4). Before both groups) to contribute to the knowledge base. In Regarding the technical background, it incorporates a same premises as our study, but which work on a basis (cf. Section 2). Then we describe the views onto the and refer to some business cases that are based on the i.e. the corporate analysts and their customers, it surwe conclude, we give a short survey of related work teresting from the application as well as from the techpasses existing systems regarding the range of support edge at the appropriate level of abstraction. Thus, much less sophisticated than CHAR. bined to support a multitude of views based on an gent access to vast amounts of small pieces of knowltheir co-workers and that allow the former (or possibly companys research department. Therefore we have to background knowledge, (v), allows a multiple view acresearch process, (iii), allows to intelligently combine Additionally, in our specic business process, we on the level of facts rather than documents, (ii), intework done in his company’s research department. He would have seen the strategic implications following of ColorBasic Inc. could have really proted from the the acquisition. Hence, the key account manager could derive a vertical integration as a possible objective for the specic user needs.
Being given this system, the key account managers grates the gathering task smoothly into the common or a statement about a general trend, about a whole the knowledge gathering and distribution process in a want our system to improve the quality and speed of set of facts, (iv), checks new facts against the available (vi), allows to route derived facts back into the comhave done the right things. assure an intuitive graphical user interface, tailored to cess to the knowledge via a single entry portal, and, from the acquisition of Grandcouleur S.A. by Bigcolor mon workplace environment. (i), supports the collective gathering of information AG. The system would have used the fact that GrandThus, what we wanted to have was a system that, facts or give a high-level view, e.g., a summarization Basic Inc. is a subsidiary of Grandcouleur S.A. and, thus, a probable supplier for Bigcolor AG in order to ments in Subsections 3.1 and 3.2. Thereby, we of parts that we have already described elsewhere the domain, an ontology. This and some further techSection 4. ([EMSS00, SS00, SAD+00]). nical assumptions will be explained in more detail in only summarize the second part as it is composed In this section we elaborate on these two requirescribed here rely on a common conceptualization of The reader may note here that the mechanisms deJ. Angele, H.-P. Schnurr, S. Staab, R. Studer 1-4 Figure 1: The entry point to CHAR 01.04.97 to how easily further development may proceed with stages of development. The main stages of the develprocess that are illustrated in Figure 6. For the main part this model is a sequential one. Nevertheless, at opment process and their key characteristics are given the design decisions that have been accomplished bein the following: each stage there is an evaluation as to whether and as fore. The results feed back into the results of earlier We distinguish dieren t phases in the development and test facts. formalized. At rst, their functionality is tested independently from the web site design in evaluIn the query development step the views and queries described in one of the earlier phases are ating those formalized queries using the ontology J. Angele, H.-P. Schnurr, S. Staab, R. Studer 1-7 4 Developing CHAR 4.1 The Development Process User requirements are collected in the requireFigure 4: A view onto strategic questions phase constitute the input for the design of the ontology. web site and for preliminary HTML pages and inments elicitation phase. Results of this very rst uence the formal domain model em bodied in the are integrated into query interface. formalized and tested during query formalization and views developed during website design, and Finally, web pages are populated, i.e. the queries Starting from the principle problem (Section 1), from tion was how to bring the required conceptual strucbased systems to corporate history analysis (Subsecplied our general methodology for developing ontologytion 4.1) including the ontology engineering part (Subissues (Subsection 4.3). swer this question, we here describe how we have aptures and reasoning capabilities into action. To anscenarios such as seen in Section 2, and from use cases section 4.2) as well as the interface and integration such as described in the previous section, the quesnew objects, attributes or relations that are not veloped. Rules describe more abstract dynamic and dynamic parts (rules) is the central issue in given explicitely. Developing the ontology conpurposes interesting concepts together with atrelations between objects which are used to derive process is started. An ontology denes a common for our case at hand in more detail in Section 4.2. taining static parts (concepts, attributes, roles) exchange knowledge about the domain. For these tied and represented. Thereafter, rules are deand layout of the web site an ontology engineering developing such an application and is described In parallel to the development of the structure language and schema used to structure, nd and tributes and relations between concepts are idention. Also right now, we are starting to test interpose types and relations between objects for annotatask, we believe, the application will always require users who are trained with the system, the domain and its underlying ontology. annotator agreement, but due to the complexity of the this static part of the ontology graphically. sents the time point. Activity has the role subject have to be modeled. There are many objects which One speciality in this domain is that time issues equate modeling primitives integrated into a logical are events in F-Logic. The attribute occursAt repretime when it is founded and there is an end time when The ontology for this application consists of (i) conmodeled. Likewise, an Acquisition is an Event which have a life time. For instance a company has a start of new knowledge. The language we use to represent occurs at a specic time, which leads to the modelmodelling of Acquisitions as Activitys which itself this company is acquired by another company or is their attributes specifying properties and relations, ing of Timepoints. The following example shows the and (iii) rules allowing inferences and the generation framework. We use our tool OntoEdit [SM00] to edit merged with another company. So durations had to be cepts dening and structuring important terms, (ii) this ontology is F-Logic [KLW95], which provides adAfter this core development process information editor etc. This process also runs in parallel to uments, by directly providing facts using the fact mentioned in Section 3.2, i.e. by annotating docis provided to the knowledge base by the tools the use of the system. quisitions and mergers to the system and the system consequences, for instance \If a company is renamed, on its own". ated, the old companies disappear, they are from now One of the key characteristics of this application was on this company has the new name", \If two companies is a central issue in this application. Modelling of time derives the consequences. For this purpose rules had to it remains the same company but from this timepoint are merged, a new company with a new name is creAll these examples show that the modelling of time rst one disapp ears at this time point and then belongs acquired by another company with share 100% then the comes part of a new company or it becomes a company that the user provides facts about new actions like acto the second one", or \If a division is outsourced it bebe modelled for all possible activities which incur such on subsidiaries of the new company", \If a company is Acquisition::Activity [ Activity::Event [subject => CompanyPart]. price =>> NUMBER; share =>> NUMBER ].
Event [occursAt => Timepoint]. hasObject =>> CompanyPart; J. Angele, H.-P. Schnurr, S. Staab, R. Studer 1-8 4.2 Ontology engineering Figure 5: The process of annotating documents which represents the company executing the activity. the role hasObject, which represents the CompanyPart Acquisition has the properties price and share and the activity applies to.
During website design the presentation strategy and In the case that a user has to click to a xed query tool. In addition to the graphical denition of queries pertext link by copy and paste from the Query Builder In the other case that a user has to ll out a form it provides output templates for the answers. These settings of HTML form elds. which then generates a query, this query has to be in F-Logic by: development of the nal presen tation of the answers to templates may then be used as starting points for the without providing further information such a query tool into the HTML editor. mation. Instead of expressing those queries directly in names N of all companies C at time t (t is between actual layouts are developed. In the query formation radio boxes or check lists, which may be used as initial F-Logic, it is more convenient to use our Query Builder phase, queries are developed which deliver the inforJavascript program. E.g. the query that returns the may be integrated directly into the web page as a hythe user. There are also templates for selection lists, start and end time of the company) may be expressed created out of the contents of the form elds by a and send it to the underlying knowledge-based system, The user has to provide time t in a form eld. The to add the url of the web page the answers are fed into form eld, has to generate the query as a string, has FORALL N,C, ST,ET <starts(C,ST) AND C:company[name->>N] AND Javascript program has to read the time t out of the ends(C,ET) AND ST <= t < ET. [DEFS99]. which in our case is the Ontobroker inference engine is well-suited to support portfolio analysis since this research services. Typically, the provided information quisitions [DIA00]. However, these databases do not haviour of competitors. Portfolio analysis [Por92] is provide any linking between the dieren t M&A actions ing access to these services at the workplace of the Corporate research itself is also an important busicollection of information about the behaviour of comIn the business community various approaches are used ent indices. E.g. The Dialog Corporation oers suc h petitors, a process that is considerably improved by is stored in databases that can be accessed via dierto analyze the development of markets and the bemethod depends on an up-to-date and comprehensive ness area. A large collection of companies oer suc h information about companies. These services can be knowledge worker. Nevertheless, these services do not facts. Reuters Business Brieng [DJR00] is oering competitors. nor do they derive additional information from these CHAR. Furthermore, the rules that are available as part of the background knowledge assist in deriving integrated into an intranet environment thus provida broad collection of services including also historical one of the most popular methods that are used. CHAR (strategic) conclusions about the behaviour of these databases addressing among others mergers and achas non-monotonic eects. For instance when a comlater acquired by another company its life time ends. erates in this region may have the strategy to expand into this region". area" or \A company which has no base in a certain pany is founded it potentially lives forever. But if it is other company which has invented this technology may possible strategical consequences had to be modelled region and which acquires another company which opby rules. For example, \A company which has no expeIn addition to the modelling of the eect of actions rience in a certain technology and which acquires anhave the strategy to become technology leader in this July 3, 2000.
Smart tools for company intelligence from the DIALOG. smart tools/compint/index.html, Observed at dialog corporation. http://library.dialog.com/ in a specic market and for providing advice about knowledge about general market behaviour. within a company. It is an intelligent assistant-type of document-based approaches do not meet the needs management system supporting the research process linked to each other and related to the background Therefore, the need to analyze the changes that ocsystem for keeping track of mergers and acquisitions means for (i) a decentralized information provisionM&A activities. More specically , CHAR provides fering an annotation tool, (ii) collecting information cur in a market increases dramatically. Conventional, CHAR (Corporate History Analyzer) is a knowledge Nowadays, markets are changing faster and faster. ing process - by using a template approach and ofstrategic implications that can be drawn from these anymore: dieren t pieces of information have to be oer deriv ed knowledge that is based on the explicitly as answers to posed queries, whereas the system outapproach to knowlege management is given. There, with respect to its integration and interlinking of inmation. For instance, the Later system for manipbe instantiated to an intelligent KM solution and how An intelligent search facility for documents is deTechnology for classifying documents and oering ing, characterizing the information sources in terms of merous publications about handling temporal inforthese commercial services the CHAR system is unique calculus used in this approach is far less powerful than ulating and querying a temporal knowledge base is sults. monotonicity is not supported by Later. described by Brusoni et al. [BCTP97]. Later oers scribed in [SSHHS98]. They use the so-called QCHAR system shows how this general approach may intelligent retrieval mechanisms. However, the Qis oering a large collection of company information temporal reasoning that includes the handling of noneither in online databases or as CDs. Compared to specied facts. In the same way Hopenstedt [Hop00] thermore, the CHAR system delivers knowledge pieces three subtasks are described: ontological engineerIn [BFG98] a general outline of an ontology-based the ontology, and intelligent knowledge retrieval. The itative and quantitative temporal relations. However, operations to assert and delete facts that specify qualConsidering the technical part, there exist, e.g., nuFrame-logic as used within the CHAR system. Furformation and the inference capabilities for deriving lined in [SSHHS98] delivers documents as query rethese three subtasks are carried out by powerful tools. new knowledge from the given facts. formation to the available background knowledge, and Future work will aim at providing better support (iv) deriving consequences that follow from these acinformation extraction. A rst step to wards this directrend analyses. These functions are built on top of tion is described in [EMSS00]. Another line of further to various kind of queries including summarizations or ous sources: a combination of linguistic and ontologythe collection of methods and tools as oered b y the into the Smart Task Support approach [SS00] thus proresearch will be the integration of the CHAR system viding assistance for researchers at their workplaces. munity Web Portal framework [SAD+00]. the knowledge base, (iii) relating these pieces of infor extracting relevant pieces of information from variquired facts and the background knowledge - by applybased methods will provide semi-automatic means for these techniques CHAR is oering integrated results ing (non-monotonic) reasoning techniques. Based on from these dieren t sources and integrating them in Ontobroker system [DEFS99] and the Semantic Com