Enhanced Count Trees In Hibernate View

Speaker(s)

Piotr Wisniewski

Affiliation

Uniwersytet Mikolaja Kopernika

Date

May 25, 2012, 12:15 p.m.

Room

room 2180

Seminar

Research Seminar of the Logic Group: Approximate reasoning in data mining

---

Counting trees are binary trees having the 
property that a node has a value representing the sum of the child nodes. The 
authors extend this concept to the trees that are not necessarily binary, in 
which each node apart the interesting source value has a value which is an 
aggregation of values of the whole subtree starting from that node. Aggregations 
can be arbitrary expressions, that can be computed from partial aggregations, 
such as SUM, COUNT, MIN, MAX, but not AVG. In the case of large hierarchical 
structures, obtaining grouping query for a group being a subtree, classically 
requires a recursive tabular expression, that is consumed by the grouping query. 
With aggregation proposed by the authors, the query reduces to reading a single 
value. The paper presents example of such an approach, then a sample interface 
for Hibernate, an object-relational mapping system, which allows you to easily 
define the relevant tables with aggregations and triggers keeping them up to 
date. The proposed interfaces retain Hibernate standards making the solution 
transparent for the programmer.

---