PDA

View Full Version : cs614 dataware house assignment no 4 solution fall 2012



rabeel
01-03-2012, 11:32 PM
Question 1: [ 08 marks ]
Compare and contrast between Episode Discovery (ED) and Context Knowledge Discovery
(CKDD) that are discussed in the research article.

Question 2: [ 12 marks ]
In this paper four context mining techniques are discussed that are: Episode Discovery (ED), Mining user Models, Context Knowledge Discovery (CKDD) and Sequential Mobile Access Pattern-mine (SMAP-mine).In your point of view which technique is best for context mining? Provide reasons to support your answer.


Note: Try to read the research paper thoroughly, read the techniques and then give your answer briefly in points form accordingly. These techniques are discussed starting from page 23 to 28, in the article attached with your assignment.

Vuhelper
01-10-2012, 07:51 PM
Compare and contrast between Episode Discovery (ED) and Context Knowledge Discovery
(CKDD) that are discussed in the research article.


This section is highly associated with pervasive event prediction approach. Context mining enables
to associate context history with current user context to increase the prediction of user action in
general with high accuracy. This section discusses four mining approaches followed by different
researchers. Episode Discovery (ED), Mining user Models, Context Knowledge Discovery
(CKDD) and Sequential Mobile Access Pattern-mine (SMAP-mine) will be discussed under this
section.




C) CONTEXT KNOWLEDGE DISCOVERY
To discuss the Context Knowledge Discovery (CKDD), let us compare it with the original KDD
(Knowledge Discovery) process. Table 2.2, discusses the difference between CKDD and KDD.
CKDD is the core function of learning and reasoning module in CAMUS [55], which includes the
steps pointed out below as reported on [51]:
i) Context data processing, which includes ontology, mining …. ;
ii) User identification (using RFID badge, PDA …) and context recognition (using
neural network, Bayesian network…);
iii) Context mining, which mines association rules, classification rule … to provide input
to learning step; and
iv) Learning, by the rule algorithm.