Abstract Semanti Web Servi es (SWS) enable the automati dis overy of distributed Web servi es based on omprehensive semanti representations.

However, although SWS te hnology supports the automati allo ation of Web

servi es for a given well-dened task, it does not entail their dis overy a or-

ding to a given situational ontext. Whereas tasks are highly dependent on

the situational ontext in whi h they o ur, SWS te hnology does not expli -

itly en ourage the representation of domain situations. Moreover, des ribing

the omplex notion of a spe i situation in all its fa ets is a ostly task and

may never rea h su ient semanti expressiveness. Parti ularly, following

the symboli SWS approa h leads to ambiguity issues and does not entail

semanti meaningfulness. Apart from that, not any real-world situation om-

pletely equals another, but has to be mat hed to a nite set of semanti ally

dened parameter des riptions to enable ontext-adaptability. To over ome

these issues, we propose Con eptual Situation Spa es (CSS) that are aligned

to established SWS standards. CSS enable the des ription of situation har-

a teristi s as members in geometri al ve tor spa es following the idea of Con-

eptual Spa es. Semanti similarity between situations is al ulated in terms

of their Eu lidean distan e within a CSS. Extending merely symboli SWS

des riptions with ontext information through CSS enables similarity-based

mat hmaking between real-world situation hara teristi s and predened re-

sour e representations as part of SWS des riptions. To prove its feasibility, we

apply our approa h to the E-Learning and E-Business domains and provide a

proof-of-on ept prototype.