ognjen.savkovic@wikispace

This page is read only. You can view the source, but not change it. Ask your administrator if you think this is wrong.
====== MAGIK Meetings Minutes ======

This document contains meeting minutes based on the discussion about the MAGIK development.


====== 17/05/2013 ======

**Attendes**: Paramita, Ognjen

**Topic**: Start of the MAGIK 3 implementation

===== Important Steps (First to Do) =====

====  Develop and Implement Controller Modul ====

      * **(First attempt)** We will create a external module (a web service) that will have a vector of connections of max size n. Once a new MAGIK instance request a db connection we use this vector as FIFO structure and provide a new connection by killing the oldest one. 
           * The problem 1 is that an instance and the controller have to communicate via object reference of new established db connection (but how?)
           * The problem 2 is that even if we have the above described mechanism how we can be sure that the object created with ''connection = DriverManager.getConnection(url, props);'' we can kill? By ''connection.close()''
      * **(Second attempt)** Use Hibernate for saving and loading on the level of schemas (//hopefully it will do all necessary job of closing unused connections for us//) - <fc #FF0000>**We try this one**</fc>

==== Hibernate ====


  * Implement //proper Hibernate// controller that saves on the level of schemas
      * Done within **(Second attempt)** above

==== Query Approximizations ====

  * Develop and implement algorithm that computes specializations of max size k for an incomplete query (TuProlog - to be investigated)
  * Develop and implement algorithm that computes generalization for an incomplete query (DLV)

==== Integrate VLDB and CIKM version ====


  * Issue 1: Reasoning with one query at the time (CIKM version) or several queries at once (as in VLDB version)
  * Issue 2: TC Suggestion at the main page or move it to the analytic page with query approximizations
      * Consult Werner

  * **First Approach** (button-up)
    * TC suggestions on main page (as before)
    * On the main page link to the analytic page 


===== Do Later =====


==== TC-statements ====


  * Constants in TC-statements should be represented with using single-quotes, **Done by Alex**
      * e.g., ''pupil('John', '1', 'b')'', 
      * in other words, instead of convention that //a constant is term that starts with a lower-case letter// it should be //a constant is a term between single quotes//
      * Note that ''a'' is different from ''"a"'' in DLV!!! (change query encoding accordingly to use double quotes)

==== Look and feel (low priority) ====
  * Representing TC-statements in a datalog form instead of the current format (consult Werner?).
      * For example,'' pupil_a(Name,Level,Code) ← pupil_i(Name,Level,Code) , learns_i(Name,'French').''
  * Implement JS tables that allow re-arraignment of table raws - up and down. (later)
  * Return coloring and style to the previous version (MAGIK CIKM)

====== 17/05/2013 (Ognjen and Werner) ======
 
Try to create a concrete task for Alex that he can work on
  * Double quotes problem?
  * Generalization algorithm?

====== 21/05/2013 ======
 
Attendees: Alex, Ognjen

Topic: Introducing double quotes in the encoding of DLV (**Ended in 28 May**)

  * Replace constants in SQL (unders single quotes) into double quotes in the CQ version of that query
  * Introduce single quotes constants into TC-statements. Then when they are parsed single quotes are replaced with the double quotes 
  * Approach: First test within the JUnit and then implement it in the code

Topic: Generalization algorithm
ognjen.savkovic@wikispace

User Tools

Site Tools

Page Tools
