A Semantic Framework for Supporting Cooperative Work in Relational Temporal Databases Paolo Terenziani, Alessio Bottrighi, Stefania Montani Dipartimento di Informatica, Univ. Piemonte Orientale, Alessandria, Italy Luca Anselma, Dipartimento di Informatica, Univ. Torino, Italy

Outline Introduction Goals and Criteria Data Model Manipulation operations Algebra Conclusions

Introduction (1/5) Cooperative work: Important, e.g. software development - Multiple alternative proposals - Selection Software engineering tools

Introduction (2/5) Cooperative work: Analogous problems using DBs to model complex domains Incremental modeling, cooperative work

* Guideline to be stored in a DB Introduction (3/5) The case of clinical guidelines: General guideline proposed by a standardization committee Proposals of update Local contextualization New therapies Evaluation of proposals * Guideline to be stored in a DB

Introduction (4/5) Open issues Augmenting DB approaches to support cooperative work, i.e.: Distinction between two phases: proposals and acceptance/rejection History of the evolution of the proposals Alternative proposals * Notice: usual semantics of (relational) DBs, conjunction of tuples

Introduction (5/5) Context Both VT and TT should be supported “Consensus” approach (TSQL2) with a high-level semantics (BCDM) BCDM supports several TDB implementations (not only TSQL2)

Goals (1/3) Extending BCDM to support cooperative updates Propose vs accept/reject Alternative proposals of updates Notice: underlined implementation

Criteria (2/3) Under-constrained policy: Super user vs user Super user operations: standard + accept/reject proposals User operations: delete (not proposals) Insert Update (chains allowed) * Notice: easy to specialize E.g.: policy 1: super users can only accept/reject

Criteria (3/3) “Minimal” extension of BCDM: Upward compatibility (manipulation operations) Reducibility (algebra)

Data Model (1/9) Two data levels needed: Super users (accepted) data User proposals * Notice: proposals need to be maintained and affect super-user data only if/when accepted

Data Model (2/9) Authoring Note: author as a data attribute - Basically a “standard” data attribute (however, author cannot be modified)

Data Model (3/9) Super user data Standard BCDM semantics

Data Model (4/9) user proposals For each super-user relation r: pi(r): set of insert proposals in r pd(r): set of proposals of deletion of tuples in r pu(r): set of updates of tuples (in r, pi(r), pu(r))

Data Model (5/9) insert proposals pi(r) is a set of standard BCDM tuples

Data Model (6/9) delete proposals pd(r) is a set of standard transaction-time tuples * Notice: no value-equivalent data in r  VT not needed

Data Model (7/9) update proposals Update involves: An origin tuple to be updated (time not needed) A new temporal tuple (standard BCDM tuple) * Notice: multiple update proposals involving the same origin are in alternative

Data Model (8/9) update proposals Definition: proposal tuple An origin A non empty set of (bi)temporal tuples t <a1,T1> <an,Tn> ……… Semantic interpretation: disjunctive set of alternative proposals (each one is a BCDM tuple)

Data Model (9/9) update proposals pu(r) is a set of proposal tuples Property: uniqueness of representation (two Proposal-relations defined over the same schema are snapshot equivalent iff they are identical )

Manipulation operations E.g.: propose update(r,origin,old,new,VT) <origin,old> identify the update proposal to be modified origin old t <a1,T1> <an,Tn> ……… IF origin=old a super-user tuple must be modified

Manipulation operations E.g.: propose update(r,origin,old,new,VT) IF admissible IF  ptpu(r) with origin(pt)=origin THEN add <origin, <new,user,UCVT>> in pu(r) IF  ptpu(r) with origin(pt)=origin  ( a1  alternatives(pt)\ a1 value equivalent to ‘new’ OR  a1  alternatives(pt)\ a1 value equivalent to ‘new’  user(a) user) THEN add ‘new’ to alternatives(pt) user(a) = user THEN add (UCVT) to the bitemporal of a1 * Notice: value equivalent proposals for the same origin are not allowed

Manipulation operations ADMISSIBILITY OF PROPOSE UPDATE OP. origin: in r or in pi(r) & current old: old (old=origin OR old origin) & current new: ( tuple t r & current & t value equivalent to ‘new’  t value equivalent to origin) &  proposal value equivalent to t with same VT

Manipulation operations ADMISSIBILITY OF PROPOSE UPDATE OP. Condition on ‘new’: example r: {<a,Ta>,<b,Tb>,…..} (r is a super-user relation) Admissible update: a  <a,T’> NOT admissible: b  <a,T’>

Manipulation operations E.g.: accept update proposal IF admissible IF  tr \ t value equivalent to origin  current(t) THEN DELETE(t); INSERT(new); close UC to all alternative proposals concerning origin IF  tr \ t value equivalent to origin  current(t)   tpi(r) \ t value equivalent to origin  current(t) THEN INSERT(new); close UC to all alternative proposals concerning origin admissible:  ptpu(r) with origin(pt)=origin  newalternatives(pt)  current(new)  [( tr \ t value equivalent to new  current(t))  t value equivalent to origin] Notice: the alternatives of the selected updated are no longer allowed

Manipulation Operations “two level” check on legal operations 1) Proposal Time Super: <a, vt1> Propose_update (x | <a, vt2>) REJECTED 2) Evaluation Time Super: <y, vt3>, <x, vt4> (1) Propose_update (y | <a, vt2>) Propose_update (x | <a, vt3>) Accept (1) Accept (2) REJECTED

Manipulation operations Property 1. Upward compatibility with BCDM Moreover, if Policy 1 is adopted: Property 2. “Semantic” upward compatibility propose(OP) accept OP Our approach BCDM

Algebraic operations Standard BCDM algebraic operations for super-user and for pi and pd Conversion operations on pu: origin(pu(r)) = {o \ pt pu(r)  o origin(pt)} = { o \ <o, (a1,…, an)>  pu(r)} alternatives(pu(r)) = {a \ ptpu(r)  a alternatives(pt)} = {(a1,…, an) \ <o, {a1,…, an}>  pu(r)}

Algebraic operations E.g.: natural join: r⋈A s = { z=<origin(z),alternatives(z)> \ IF $pt1Îr , $pt2Îs \ origin(pt1)[A]= origin(pt2) [A] Ù $a1Îalternatives(pt1), $a2Îalternatives(pt2) \ a1[A]=a2[A] Ù a1[T]a2[T]  THEN origin(z)[A]=origin(pt1)[A] Ù z[B]=origin(pt1)[B] Ù z[C]=origin(pt2)[C] Ù altÎalternatives(z), where alt[A]=a1[A]=a2[A] Ù alt[B]=a1[B] Ù alt[C]=a2[C] Ù alt[T]=a1[T]a2[T] }

Algebraic operations Definition: conv conv(pu(r))={(a1,…,an,a’1,…,a’n,T)\ ptpu(r) \ (a1,…,an)=origin(pt)  (a’1,…,a’n)=alternatives(pt) } t <a1,T1> <an,Tn> ……… Semantic level Tn t an T1 a1 … Relational level conv

Algebraic operations Property: reducibility (!?) conv( OpA( pu(r) ) ) = OpBCDM( conv( pu(r) ) ) * Note: underlying possible implementation

Implementation (idea) (Data Abstraction) SEMANTIC Level PROPOSAL RELATION Conv Accept Op Propose Op Algebraic Op Accept Op Propose Op Algebraic Op

Conclusions Problem of cooperative update to DB’s is important New problem in DB field Semantic approach extending BCDM to support (1) proposal\evaluation & (2) alternative proposals Data model Manipulation operations Algebra Upward compatibility\reducibility Easy Implementability