The Corporate History Analyzer
Jurgen Angele b
,Hans-Peter Schnurr a,b
, Steen Staab a,b
,Rudi Studer a,b
E-Mail: fangele,schnurr,staab,studer g@ont opri se.d e
a
InstituteAIFB, University of Karlsruhe, 76128 Karlsruhe
http://www.aifb.uni-karl sruh e.de /WB S
b
Ontoprise GmbH,Haid-und-Neu Strae7, 76131 Karlsruhe,Germany
http://www.ontoprise.com
Comegather'roundpeople
Whereveryouroam
Andadmitthatthewaters
Aroundyouhavegrown
Andacceptitthatsoon
You'llbedrenchedtothebone.
Ifyourtimetoyou
Isworthsavin'
Thenyoubetterstartswimmin'
Oryou'llsinklikeastone
Forthetimestheyarea-changin'.
BobDylan,1963
Abstract
Up-to-date corporate information research,
i.e. the active tracking and management of
knowledgerelevanttoone'sbusiness,isama-
jor task for knowledge-intensive companies.
While the correct analysis of market situ-
ations and competitors are critical require-
mentsfor success, thefailure to provide ade-
quateknowledgeaboutone'sbusinessenviron-
mentmayincurlargelosses. Inorderto sup-
port thetrackingof externalbusiness knowl-
edge and its analysis, we have built CHAR,
theCorporate History AnalyseR,anew type
of system for gathering and analysing tem-
poral data about companies. CHAR inte-
grates facts about organizational structures
andstrategicactivitiesintoacommonknowl-
edge base. Building on an ontology with
an elaborate temporal structure, knowledge
Thecopyrightofthispaperbelongstothepaper'sauthors. Per-
missiontocopywithoutfeeallorpartofthismaterialisgranted
provided thatthecopies arenotmade ordistributed fordirect
commercialadvantage.
Proc.oftheThirdInt.Conf. onPracticalAspects
ofKnowledge Management(PAKM2000)
Basel,Switzerland,30-31 Oct.2000,(U.Reimered.)
http://sunsite.informatik.rwth-aachen.de/Publications/CEUR-WS/Vol-34/
pieces are amalgamated in order to derive a
plentitudeofsummarizingviews ontocurrent
andpreviousstatesofaairs.
1 Introduction
Global marketschangingwith aneverfaster paceare
oneof the key topics in business nowadays. Keeping
trackofwhatchangesinthemarketsandrecognizing
whatmovesone'scompetitorsisvitaltoanyenterprise
thatwantstosucceedonaglobalscale. Whilethishas
beenatruism formanagementat thetopmost strate-
gic level for a long time, the very same necessity is
currentlyspringinguponallmanagementlevelsdown
totheexpertknowledgeworkerwhomanageshisown
skills.
The trouble is that these management levels and
professionals do haveahard time gatheringinforma-
tion, analyzing it, and performing their operational
work,suchasresearchinganddevelopingnewproducts
or streamlining their production lines. This means
thattheevaluationofwhathappensinone'sownbusi-
nessmayoftenbeneglected.
Here comescorporate researchinto play. Thetask
of the corporate researchers is the tracking of rele-
vantknowledgeintheoutsidebusinesssettingandthe
communicationofimportantknowledgeto stakehold-
erswithin thecompany. The generaldiÆculty, then,
isthedeliveryofknowledgetotherightpersonsatthe
righttime and atthe right level ofabstraction. Peo-
pledoneitherwanttobeoodedwithfactsthat they
do not need nor do they want to be swamped with
nittygrittydetails,buttheyratherneedasummariz-
ingviewontothesinglepiecesofknowledge.
Withtheseaimsinmind,wehavedevelopedCHAR,
theCorporate History Analyzer, asystemthat allows
for collectivegatheringof information and forintelli-
edge at the appropriate level of abstraction. Thus,
CHAR constitutes an new typeof systemthat is in-
terestingfromtheapplicationaswellasfromthetech-
nical point of view. Seen from theside of theusers,
i.e. thecorporateanalystsandtheircustomers,itsur-
passesexistingsystemsregardingtherangeofsupport
itoersonthequeryingandontheprovisioningside.
Regardingthetechnicalbackground,itincorporatesa
rangeofontology-basedknowledgemanagementtools
thatareinterestinginthemselvesandwhicharecom-
bined to support a multitude of views based on an
ontologywithanelaboratetemporalstructure.
Intherest of thispaperwewillrst elaborate the
businesscasethatthehistoryanalyzermaybeusedfor
(cf. Section 2). Then we describethe views onto the
knowledgebasethatsupportcorporateresearchersand
theirco-workersandthatallowtheformer(orpossibly
bothgroups)to contributetotheknowledgebase. In
a nal step we describe how the overall system has
beenconceivedand implemented (Section 4). Before
we conclude, we give a short survey of related work
andrefertosomebusinesscasesthatarebasedonthe
samepremisesasourstudy,butwhichworkonabasis
muchlesssophisticatedthanCHAR.
2 A Case Study | Restructuring in
the Chemical Industry
ConsiderthecaseyouarekeyaccountmanagerofCol-
orBasic Inc, asupplierof basicmaterial forthe color
industry, responsible for several key accounts in the
EuropeancolorIndustry. Theotherdayyoureadina
newspaperarticle,thatoneofyourbiggestcustomers,
the German BigcolorAG is in acquisition talks with
theirfrenchcompetitorGrandcouleurS.A.
So what? Ina rst happyreactionaboutthis de-
velopment,youdiscusswithyourproductionmanager
theabilityofyourcompanytoincreasetheproduction
of basicmaterialin ashort timeperiod. Youplan to
convinceyourkeyaccountatBigcolorAGtoenhance
your supply agreement also for their new subsidiary
immediatelyafteraquisition. Niceanalysis! Yourpro-
duction manager conrms a possible increase of pro-
ductionandpreparesaquickswitchofproductionline.
Twomonthlater,youreadthenewsabouttheacquisi-
tionofGrandcouleurS.A.and immediately,youmeet
yourkeyaccountatBigcolorAGtogethisagreement
forhighersalesvolumeofbasicmaterialwithveryex-
ibledeliveryconditions. Nicestrategy!
Yourkeyaccountisimpressedaboutyourquickre-
action and your fast and exible supply conditions,
buttellsyou,thatBigcolorAGacquiredGrandcouleur
S.A.duetothefactthattheyhaveasubsidiaryinthe
basicmaterialindustry,GrandBasicInc.,yourbiggest
integration and bought very interesting supply con-
ditionsviatheacquistionofGrandcouleurS.A.Thus,
unfortunately,insteadofhighersalesvolumeyouwont
haveanysales in near future with BigcolorAG. Bad
News!
Now, it is easy to recognize, that your analysis of
the situation waswrong. You had notenough infor-
mationabouttheacquisitionto createabetterstrat-
egy. Later on,youwill hearalotofcolleagues telling
youstoriesaboutthewell-knownrelationshipbetween
Grandcouleur S.A.and GrandBasicInc. Evenworse,
but howshouldyouhandle such situations? There is
aneverfasterpaceofchangeinglobalmarkets. There
is no day without news about mergers, acquisitions
and alot of otherstrategic movesof playersin every
industry. Oneofthemostdynamicmarketsinthepast
yearshasbeenthechemicalindustry. Here,bigglobal
playerspermanentlyrestructuretheirorganization,try
toacquiresmallinnovativetechnologyleadersorght
aboutentrystrategiesin newmarkets. This verydy-
namicsituation leadstotheproblem, thatnomarket
participant has a permanent clear picture about the
industry.
Nowadays,themarketanalysts,consultantsandin-
house market research departments try to track the
activities of their industry with traditional methods.
Newspaperarticles,onlinedatabasesandannualcom-
panyreportswereanalysedandcompetitorswebpages
werethoroughlytracked. Theresultsarepresentedto
themanagement,publishedinreportsanddistributed.
Thereexist severalproblemsinthis researchprocess:
Information archivesare document-based. For a
collectivegatheringoffactsthisviewistoocoarse.
Bythis way onemay introducean abundanceof
redundancies on the one hand and one may ac-
countfor onlyfewofmanyaspects thatoccurin
aparticulardocument(e.g. onlytheinformation
given in the headlines), thus missing the crucial
piecesofinformationagain.
Thenextproblem stemsfrom thefact thattypi-
cally documentmanagementsystemsrelyalmost
exclusively on information retrieval techniques,
whichareratherunreliable.
Even if convential annotation is used for these
documents, it is most often too cumbersome to
describe the relevant implications, too. E.g., if
one company sells one of its units, the implica-
tion is that this unit may no longer be the sole
supplierfortheoverallcompanyasitwasbefore.
Theseimplicationsmayonlybemadetransparent
if background knowledge is used, but document
thesauri|themostadvancedones employtopic
maps.
Finally, dierent executives need dierent views
ontothesamebasicpiecesofinformation. What
is relevant to the oneemployee, e.g., acompeti-
tor's expansion in his regional market, may be
irrelevantfromamoreglobalpointofview.
Thus, what wewantedto havewasasystem that,
(i), supports the collective gathering of information
ontheleveloffacts ratherthandocuments,(ii), inte-
grates the gathering tasksmoothlyinto thecommon
research process, (iii),allowsto intelligently combine
facts or givea high-level view, e.g.,a summarization
ora statement abouta generaltrend, about awhole
setoffacts,(iv),checksnewfactsagainsttheavailable
backgroundknowledge,(v),allowsamultipleviewac-
cess to the knowledge via asingle entry portal, and,
(vi), allowsto route derived facts backinto the com-
monworkplaceenvironment.
Additionally, in our specic business process, we
wantoursystem toimprovethequalityandspeedof
theknowledgegatheringanddistributionprocessin a
companysresearchdepartment. Thereforewehaveto
assureanintuitivegraphicaluserinterface,tailoredto
thespecicuserneeds.
Beinggiventhissystem,thekeyaccountmanagers
ofColorBasicInc. couldhavereallyprotedfrom the
workdonein his company'sresearchdepartment. He
would have seen the strategic implications following
fromtheacquisitionofGrandcouleurS.A.byBigcolor
AG.ThesystemwouldhaveusedthefactthatGrand-
Basic Inc. is asubsidiary of Grandcouleur S.A. and,
thus, aprobablesupplierforBigcolorAGin order to
deriveaverticalintegrationasapossibleobjectivefor
theacquisition.Hence,thekeyaccountmanagercould
havedonetherightthings.
3 CHAR | The system
Intheprevioussectionwehaveseensomepitfallsthat
may appear while tracking one's markets and com-
petitors. From these problems we have derived the
requirementsthatasystemforcorporatehistorytrack-
ingshould meet:
1. It should provide multiple views onto the
same knowledge base for dierent time frames,
for dierent regional foci, for varying intra-
organisationalstructuresandfor dierentstrate-
gicquestions,tonamebutafew.
2. It should allow many people to contribute fac-
tual knowledge in a way that is embedded into
aroundasemanticbackground.
In this section we elaborate on these two require-
ments in Subsections 3.1 and 3.2. Thereby, we
only summarize the second part as it is composed
of parts that we have already described elsewhere
([EMSS00,SS00,SAD +
00]).
Thereadermaynoteherethatthemechanismsde-
scribed here rely on a common conceptualization of
thedomain,anontology. Thisandsomefurthertech-
nical assumptionswill beexplained in moredetail in
Section 4.
3.1 Querying Knowledge
ThequeryinterfaceoftheCorporateHistoryAnalyzer
has been developed in order to deal with organiza-
tional and strategicquestions that depend onspatio-
temporal constraints. It renders views that may be
seen ona commonweb browser. Actually, they look
just like common web sites with some choices for se-
lection. These selections are also controlled by the
knowledge of the backend system, e.g. one can only
selectcompaniesthatareknowntoexistintheknowl-
edgerepository.
Figure 1 depicts the main views that are oered
by CHAR, viz. Activities, Organization, Know How,
Strategic Questions and General Query Possibilities
(indicated by \Search"). While the latter may be
asked by means such as known from standard tools
for OLAP/business intelligence, we consider to help
the userwith the corporate historyanalyzer through
views that directly reecthis interestsontheorgani-
zationalandstrategiclevel.
3.1.1 Answer organizationalquestions
Therstmajorcategoryofqueriesrelevanttothecor-
porate historyis aboutorganizationalstructures and
the activities that change organizational structures.
Forinstance,theviewof\acquisitionsofM.A.Hanna"
returnsallitspurchases(cf.Fig.2),andcorresponding
viewsareoeredforSellings,Mergers,Restructurings
andManagementChanges(Figure1).
Whatisinterestingtonoteatthispointisthatitis
rather diÆculttoget aclear pictureof what isreally
happeningwith M.A.Hanna. Itis diÆcultandtime-
consumingforthehumananalysttodetectsometrend
inlistsofsinglefacts. Observationsbecomemucheas-
ier, when dierent typesof facts may berelated and
contrasted. For instance, Figure 3depicts twosnap-
shots of M.A.Hanna's organigramthat areautomat-
ically derived from single activities, like acquisitions
and restructurings, and that give the analyst a neat
picture ofhow formerlyisolated purchases that M.A.
inthecompanyin1997(e.g.,\CompoundingTechnol-
ogy"havingbeenreorganizedintothe BusinessArea
\PlasticCompounding").
It shouldhavebecome clearthatthis typeof com-
prehensivesupport isbeyondthescopeofstraightfor-
warddatabasesolutions,butratherrequiresextensive
programmingorthetypeof mechanismsthatwerely
onforCHAR|describedin detailinSection 4.
3.1.2 Answer strategic questions
This far, we have described how CHAR provides so-
phisticated supportthatisbasedon\hardfacts"and
that does not aim at interpretation of single facts.
With itssupport for strategic questions,CHAR tries
to give answersto questions aboutbusiness competi-
tors which cannot be answereddenitely, but which
rely onsomeconjectures. For instance, thepurchase
ofacompanyfromabroadmayleadtoagainofmar-
ket sharein that area,andthusto aregionaldiversi-
cation. Thus, the rst question in in Figure 4 is a
prototypestrategicquestion. Theoveralllistthat we
currentlysupport readsas:
1. Which Activityof Xleadstooperation inregion
Y?
2. Is there any regional expansionof X due to any
acquisitionsinceT?
3. Is there any regional expansionof X due to the
acquisitionofY?
4. Whatactivityof XmodiedtheirBusinessArea
Y?
5. WhatexperiencegainsX fromacquisitionofY?
Answersfor someof these questions are presented
in thelower partof Figure 4. Forexample, Hanna's
acquisition of Bergmann in 1994 led to anew opera-
tionin WesternEurope|exploiting thebackground
knowledgethatBergmann'sGermanbaseliesinWest-
ern Europe(lower left of Figure 4). Question 2trig-
gersasearchforanykindofregionalexpansionwithin
some time frame, which may be answered positively
withthecaseofEnviroCareCompounds(middle),Fi-
nally, Hanna's business areaof PlasticCompounding
wasaected byrestructurings such astheone of the
companypartthatformerlywasBergmann.
By now it should be obviousthat for this type of
questionsoneneedscomprehensivereasoningsupport,
suchasrulesaboutgeographicrelationships(Japanis
inAsia,ItalyisinSouthernEurope,etc.),rulesabout
categorization of know how, as well as rules about
strategic indications (purchase may be indicative for
regionalexpansion).
Finally,theinformationseekermayalsouse ourtree-
basedandhyperbolicinterfaces(cf.[SAD +
00])tocom-
pile\arbitrary"questions. These questionsmaycom-
prise conceptual and relationalrestrictions according
tothesystem'sontologyand/orregularexpressions.
3.2 Providing Knowledge
The process of providing new facts into the knowl-
edge warehouse should be as easy and as smoothly
integratedintothecommonworkingtasksaspossible.
Forthisreasonweoervariousmodesofcontributing
knowledge.
First, one mayenter information through aform-
basedinterface that isextended byacanonicaland a
hyperbolic tree view onto the concept hierarchy (cf.
[SAD +
00]).
Second, when the information that is to be pro-
vided is produced during the writing of documenta-
tions or reports onemayuse atemplate approach in
order to generate knowledge by writing these papers
(cf.[SS00]).
Third, one may use wrapper mechanismsin order
to providedatafrom tablesand listsontheweb. For
instance, the approach described in [BD99] uses the
same basicrepresentation (viz. F-Logic, cf. next sec-
tion)asitssemanticbackboneandmay,thus,directly
feedinto ourknowledgebase.
Fourth|andmostimportantforCHAR,onemay
use our annotationtool in order to add metadatato
data given in documents(cf. [EMSS00]). A snapshot
oftheannotationtoolnotingsomeactionaboutM.A.
Hannaisshownin Figure5: Theuserreadsorworks
with documentsusing atext orspreadsheet process-
ingtooloraninternetbrowser. Whenhedetectssome
relevantchange beingdescribedin thedocumentand
thischangemightbecomerelevant 1
,hehighlightsthe
wordor phraseandusestheannotationtooltoselect
the typeof the phrase(e.g. \Hanna M.A." isa com-
pany) and its relation to other material (e.g. Hanna
sells ShapesDistribution Business to GE Plastics on
May11, 2000). Thedocument,these factsand meta-
dataabouttheannotater,thetimeofannotation,etc.
isall storedin theback-enddatabase. Currentlythis
is apurely manual process. Inthe future weplan to
useinformationretrievalmethodstorecognizephrases
(or words) that had been seenand annotatedbefore
(e.g., if there is a second document about \Hanna
M.A.")andinformationextractionmethodsthat pro-
1
Our rst intuition was that every major organizational
change of one'scompetitorsor clientsshouldbenotedbythe
company'sITresearchdepartment. Overtimethisroughesti-
mationwillhavetobespeciedmoreprecisely|alsodepending
onthetypeofchangeanditsnancialimplication.
01.10.93
01.04.97
Figure3: Beforeandafterrestructurings
pose typesand relations between objects forannota-
tion. Also right now, we are starting to test inter-
annotatoragreement,butduetothecomplexityofthe
task, we believe, the application will always require
users who are trained with the system, the domain
anditsunderlyingontology.
4 Developing CHAR
Starting fromtheprincipleproblem(Section1), from
scenariossuchasseeninSection2,andfromusecases
such as described in the previous section, the ques-
tion was howto bring the requiredconceptual struc-
tures and reasoning capabilities into action. To an-
swerthis question,we heredescribehowwehaveap-
pliedourgeneralmethodologyfordevelopingontology-
based systems to corporate history analysis (Subsec-
tion4.1)includingtheontologyengineeringpart(Sub-
section 4.2) as well as the interface and integration
issues(Subsection4.3).
4.1 The DevelopmentProcess
We distinguish dierent phases in the development
process thatareillustratedin Figure6. Forthemain
part this model is a sequential one. Nevertheless, at
eachstagethereisanevaluationastowhetherandas
to howeasily further development may proceed with
thedesign decisionsthathavebeenaccomplishedbe-
fore. The resultsfeed back into the results of earlier
stages ofdevelopment. Themain stagesof thedevel-
opmentprocessandtheirkeycharacteristicsaregiven
in thefollowing:
User requirements are collected in the require-
mentselicitationphase. Resultsof thisveryrst
phase constitute the input for the design of the
websiteandforpreliminaryHTMLpagesandin-
uencetheformaldomainmodelembodiedinthe
ontology.
In parallel to the development of the structure
andlayoutofthewebsiteanontologyengineering
processisstarted. Anontologydenesacommon
languageandschema usedto structure, ndand
exchangeknowledgeaboutthedomain. Forthese
purposes interesting concepts together with at-
tributesandrelationsbetweenconceptsareiden-
tied and represented. Thereafter, rules are de-
veloped. Rules describe more abstract dynamic
relationsbetweenobjectswhichareusedtoderive
new objects, attributes orrelations that are not
given explicitely. Developing the ontology con-
taining static parts (concepts, attributes, roles)
and dynamic parts (rules) is thecentral issue in
developing such an application and is described
forourcaseathandinmoredetailinSection4.2.
In the query development step the views and
queriesdescribedin oneof theearlierphasesare
formalized. At rst, theirfunctionality is tested
independentlyfrom the website design in evalu-
atingthoseformalizedqueriesusingtheontology
andtestfacts.
Finally,webpagesarepopulated,i.e. thequeries
and views developed during website design, and
formalizedandtestedduring queryformalization
areintegratedintoqueryinterface.
After this core development process information
is provided to the knowledge base by the tools
mentionedinSection 3.2,i.e. byannotatingdoc-
uments,bydirectlyprovidingfactsusingthefact
editor etc. This process also runs in parallel to
theuseofthesystem.
4.2 Ontology engineering
The ontologyfor this application consists of (i) con-
cepts dening and structuring important terms, (ii)
their attributes specifying properties and relations,
and (iii) rules allowing inferencesand the generation
of newknowledge. The languageweuse to represent
this ontologyis F-Logic[KLW95],whichprovidesad-
equate modeling primitives integrated into a logical
framework. WeuseourtoolOntoEdit[SM00] toedit
thisstaticpartof theontologygraphically.
One speciality in this domain is that time issues
have to be modeled. There are many objectswhich
havea life time. Forinstance acompany hasastart
timewhenitisfoundedandthereisanendtimewhen
this company is acquired by another company or is
mergedwithanothercompany. Sodurationshadtobe
modeled. Likewise,anAcquisitionisanEventwhich
occurs at a specic time, which leads to the model-
ing of Timepoints. Thefollowingexampleshows the
modellingof AcquisitionsasActivityswhichitself
areeventsin F-Logic. Theattribute occursAtrepre-
sentsthetime point. Activityhas therolesubject
which representsthecompanyexecuting theactivity.
Acquisitionhasthepropertiespriceandshareand
therolehasObject,whichrepresentstheCompanyPart
theactivityappliesto.
Event [occursAt => Timepoint].
Activity::Event [subject => CompanyPart].
Acquisition::Activity [
hasObject =>> CompanyPart;
price =>> NUMBER;
share =>> NUMBER ].
Oneofthekeycharacteristicsofthisapplicationwas
thattheuserprovidesfactsaboutnewactionslikeac-
quisitions and mergersto the systemand the system
derivestheconsequences. Forthispurposeruleshadto
bemodelledforallpossibleactivitieswhichincursuch
consequences,forinstance \If acompany isrenamed,
it remains the same company but from this timepoint
onthiscompanyhasthenewname",\Iftwocompanies
are merged, anew company with a new name is cre-
ated, the old companies disappear, they arefromnow
onsubsidiariesofthenewcompany",\Ifacompanyis
acquiredbyanothercompanywithshare100%thenthe
rstonedisappearsatthistimepointandthenbelongs
tothesecondone",or\Ifadivisionisoutsourceditbe-
comespartofanewcompanyoritbecomesacompany
on itsown".
Alltheseexamplesshowthatthemodellingoftime
isacentralissueinthisapplication. Modellingoftime
hasnon-monotonic eects. Forinstancewhen acom-
panyisfoundeditpotentiallylivesforever. Butifitis
lateracquiredbyanothercompanyitslifetimeends.
Inadditiontothemodellingoftheeect ofactions
possiblestrategical consequenceshad to be modelled
byrules. Forexample,\Acompanywhichhasnoexpe-
rience in acertaintechnology and which acquires an-
othercompanywhichhas inventedthistechnologymay
have the strategy to become technology leader in this
area" or\A company which has no basein acertain
region andwhich acquiresanother company which op-
eratesin this region may have the strategy to expand
intothis region".
4.3 Glueing allparts together
During website design the presentation strategy and
actual layoutsare developed. Inthe queryformation
phase, queries are developed which deliver the infor-
mation. Insteadofexpressingthosequeriesdirectlyin
F-Logic,itismoreconvenienttouseourQueryBuilder
tool. Inadditiontothegraphicaldenitionofqueries
it provides output templates for the answers. These
templatesmaythenbeusedasstartingpointsforthe
developmentofthenalpresentationoftheanswersto
the user. There arealso templates for selectionlists,
radioboxesorchecklists,whichmaybeusedasinitial
settingsofHTMLform elds.
Inthecasethatauserhastoclicktoaxedquery
without providing further information such a query
maybeintegrateddirectlyinto thewebpageasahy-
pertextlinkbycopyandpastefromtheQueryBuilder
toolinto theHTMLeditor.
Inthe othercasethat auserhasto lloutaform
which then generates a query, this query has to be
created out of the contents of the form elds by a
Javascript program. E.g. thequery that returns the
names N of all companies C at time t (t is between
startandendtimeofthecompany)maybeexpressed
in F-Logicby:
FORALL N,C, ST,ET <-
C:company[name->>N] AND
starts(C,ST) AND
ends(C,ET) AND
ST <= t < ET.
Theuserhastoprovidetimetinaformeld. The
Javascript programhas to read thetime t outof the
form eld, hasto generate thequery asastring, has
toaddtheurlofthewebpagetheanswersarefedinto
andsendittotheunderlyingknowledge-basedsystem,
which in our case is the Ontobrokerinference engine
[DEFS99].
5 Related Work
Inthebusinesscommunityvariousapproachesareused
to analyze the development of markets and the be-
haviour of competitors. Portfolio analysis [Por92] is
oneofthemostpopularmethodsthatareused. CHAR
is well-suited to support portfolio analysis since this
method depends onanup-to-dateandcomprehensive
collectionofinformationaboutthebehaviourofcom-
petitors, a process that is considerably improved by
CHAR. Furthermore, the rules that are available as
part of the background knowledge assist in deriving
(strategic) conclusions about the behaviour of these
competitors.
Corporateresearchitselfisalsoanimportantbusi-
ness area. A large collectionof companies oersuch
researchservices. Typically,theprovidedinformation
isstored in databasesthat canbeaccessed viadier-
entindices. E.g. TheDialog Corporationoerssuch
databases addressing among others mergers and ac-
quisitions [DIA00]. However, these databasesdo not
provideanylinkingbetweenthedierentM&Aactions
nor dothey derive additionalinformation from these
facts. Reuters Business Brieng [DJR00] is oering
abroadcollectionofservicesincluding alsohistorical
information about companies. These servicescan be
integrated into an intranet environment thus provid-
ing access to these services at the workplace of the
knowledgeworker. Nevertheless,theseservicesdonot
speciedfacts. Inthe samewayHopenstedt [Hop00]
is oering a large collection of company information
either in online databases or as CDs. Compared to
thesecommercialservicestheCHARsystemisunique
with respect to its integration and interlinking of in-
formation and the inference capabilities for deriving
newknowledgefromthegivenfacts.
Consideringthetechnicalpart,thereexist,e.g.,nu-
merous publications about handling temporal infor-
mation. For instance, the Later system for manip-
ulating and querying a temporal knowledge base is
described by Brusoni et al. [BCTP97]. Later oers
operationstoassertanddeletefactsthatspecifyqual-
itativeandquantitativetemporalrelations. However,
temporalreasoningthatincludesthehandlingofnon-
monotonicityisnotsupported byLater.
In [BFG98] a generaloutline ofan ontology-based
approach to knowlege management is given. There,
three subtasks are described: ontological engineer-
ing,characterizingtheinformationsourcesintermsof
theontology,andintelligentknowledgeretrieval. The
CHAR system showshowthis general approach may
beinstantiatedtoanintelligentKMsolutionandhow
thesethreesubtasksarecarriedoutbypowerfultools.
An intelligent search facility for documents is de-
scribed in [SSHHS98]. They use the so-called Q-
Technology for classifying documents and oering
intelligent retrieval mechanisms. However, the Q-
calculususedinthisapproachisfarlesspowerfulthan
Frame-logicas used within the CHAR system. Fur-
thermore,theCHARsystemdeliversknowledgepieces
asanswersto posed queries,whereasthe systemout-
lined in [SSHHS98] delivers documents as query re-
sults.
6 Conclusion
Nowadays, markets are changing faster and faster.
Therefore, the need to analyze the changes that oc-
curin amarketincreasesdramatically. Conventional,
document-based approaches do not meet the needs
anymore: dierent pieces of information have to be
linked to each other and related to the background
knowledgeaboutgeneralmarketbehaviour.
CHAR(CorporateHistoryAnalyzer)isaknowledge
management system supporting the research process
withinacompany. Itisanintelligentassistant-typeof
system for keepingtrack of mergers and acquisitions
in a specic market and for providing advice about
strategic implications that can be drawn from these
M&A activities. More specically, CHAR provides
means for (i) a decentralized information provision-
ing process - by using a template approach and of-
fering an annotation tool, (ii) collecting information
the knowledge base, (iii) relating these pieces of in-
formationtotheavailablebackgroundknowledge,and
(iv) deriving consequences that followfrom these ac-
quiredfactsandthebackgroundknowledge-byapply-
ing (non-monotonic) reasoningtechniques. Basedon
these techniquesCHAR is oering integrated results
tovariouskindofqueriesincludingsummarizationsor
trend analyses. These functions are built on top of
the collectionof methods and toolsas oered by the
Ontobrokersystem[DEFS99]andtheSemanticCom-
munityWebPortalframework[SAD +
00].
Future work will aim at providing better support
forextractingrelevantpiecesofinformationfromvari-
oussources: acombinationoflinguisticandontology-
basedmethodswillprovidesemi-automaticmeansfor
informationextraction. Arststeptowardsthisdirec-
tionisdescribedin[EMSS00]. Anotherlineoffurther
research will bethe integrationof the CHAR system
intotheSmartTaskSupportapproach[SS00]thuspro-
vidingassistance forresearchersattheirworkplaces.
References
[BCTP97] V. Brusoni, L. Console, P. Terenziani, and
B.Pernici. Later: Managing temporal infor-
mationeÆciently. IEEE Expert, pages56{63,
July/August1997.
[BD99] M. Bauer and D. Dengler. Trias: Trainable
informationassistantsforcooperativeproblem
solving. InProceedings ofAgents-1999, pages
260{267.ACMPress,1999.
[BFG98] V. R. Benjamins, D. Fensel, and A. Gomez.
Knowledge management through ontologies.
InReimer[Rei98 ].
[DEFS99] S. Decker, M. Erdmann, D. Fensel, and
R. Studer. Ontobroker: Ontology Based Ac-
cess to Distributed and Semi-Structured In-
formation. In R. Meersman et al., editors,
Database Semantics: Semantic IssuesinMul-
timediaSystems, pages 351{369.KluwerAca-
demicPublisher,1999.
[DIA00] DIALOG.
Smarttoolsforcompanyintelligencefromthe
dialogcorporation. http://library.dialog.com/
smarttools/compint/index.html,Observedat
July3,2000.
[DJR00] DowJonesReutersBusiness InteractiveLLC.
Trading as factiva. factiva, a Dow Jones
Reuters company. http://www.factiva.com/
rbb/companies.asp?node=rbb-link4,
ObservedatJuly3,2000.
[EMSS00] M.Erdmann,A.Maedche,H.-P.Schnurr,and
S. Staab. From manual to semi-automatic
semantic annotation: About ontology-based
text annotation tools. In P. Buitelaar and
ING2000 WorkshoponSemanticAnnotation
and Intelligent Content. Luxembourg, August
5-6,2000,2000.
[Hop00] Hoppenstedt. HoppenstedtVerlagsprogramm
Firmendatenbank.
http://www.hoppenstedt.de/ hfdb1.asp,Ob-
servedatJuly3,2000.
[KLW95] M. Kifer, G. Lausen, and J. Wu. Logi-
calfoundations of object-oriented and frame-
based languages. Journal of the ACM, 42,
1995.
[Por92] M. E. Porter. Wettbewerbsstrategie (Compet-
itive Strategy). Campus Verlag, Frankfurt,
1992.
[Rei98] U.Reimer,editor.PAKM98PracticalAspects
of Knowledge Management | Proceedings of
theSecondInternationalConference,1998.
[SAD +
00] S.Staab, J. Angele,S. Decker, M. Erdmann,
A.Hotho,A.Maedche,R.Studer,andY.Sure.
SemanticCommunityWebPortals.InProceed-
ings of the 9th World Wide Web Conference
(WWW-9),Amsterdam,Netherlands,2000.
[SM00] S.StaabandM.Maedche.Ontologyengineer-
ingbeyondthemodelingofconceptsandrela-
tions. InProceedingsoftheECAI2000Work-
shoponOntologiesandProblem-solvingMeth-
ods.Berlin,August2000,2000.
[SS00] S.Staaband H.-P.Schnurr. Smarttask sup-
portthroughproactiveaccesstoorganizational
memory.JournalofKnowledge-basedSystems,
September2000.
[SSHHS98] K. Stanoevska-Slabeva, A. Hombrecher,
S. Handschuh, and B. Schmid. EÆcient in-
formationretrieval: Toolsforknowledgeman-
agement. InReimer[Rei98 ].