Semantic matchmaking and filtering Clause Samples

Semantic matchmaking and filtering. Match-making by the ▇▇▇▇▇▇▇▇ takes as input the semantic knowledge base for each problem, expressed in a variant of the KRSS6 ontological notation, i.e., in the case of match-making that is the merge of – the reference knowledge base (LUMO, cf. also D 4.4), – the user profile (cf. D 4.4), and – the content annotation for a single content item from the list of candidate items that are available (cf. D 2.4) . A manufactured concept – for demonstration purposes here called UIC (after User Interests Concept) – is included in the profile, for which it applies: EhasInterest.(AHBHC H...) ± UIC (9) where A,B,C are concepts that denote user interests. These concepts can either be LUMO classes, individuals instantiating a LUMO class, or complex concepts, i.e., logical expressions, for example: ER.(DHE) ± C (10) where R is a LUMO property, D and E are LUMO classes. Similarly, another axiom exists for disinterest, implying a manufactured concept UDC (after User Disinterest Concept), for which it holds that it is disjoint with UIC, such that: UIC H UDC ±⊥ (11) UIC and UDC being disjoint, enabling the ▇▇▇▇▇▇▇▇ to refute a match whenever UDC is satisfied (see below), even if UIC is also satisfied.‌ 3.2.1.1 Content Filtering The purpose of match-making (i.e. content filtering) is to determine whether UIC is satisfied by a given content item. For this purpose, the ▇▇▇▇▇▇▇▇ runs once per content item in a list of candidate content items. If UIC is satisfied then the candidate matches the user profile and accordingly an entry of the match is stored internally until all candidates are examined. 3.2.1.2 Toy Example Let us assume a user from Berlin who is interested in sport events, but only those that are in the user’s local vicinity. The user also bares an interest in politics in general. However, he has shown specific disinterest in ▇▇▇▇▇▇ ▇▇▇▇▇ whom he dislikes, by rejecting news items about him, either regarding politics or personal news stories. The user model would read: