Mining Rank Data

Abstract

The problem of frequent pattern mining has been studied quite extensively for various types of data, including sets, sequences, and graphs. Somewhat surprisingly, another important type of data, namely rank data, has received very little attention in data mining so far. In this paper, we therefore addresses the problem of mining rank data, that is, data in the form of rankings (total orders) of an underlying set of items. More specifically, two types of patterns are considered, namely frequent rankings and dependencies between such rankings in the form of association rules. Algorithms for mining frequent rankings and frequent closed rankings are proposed and tested experimentally, using both synthetic and real data. arXiv:1806.05897v1 [cs.LG] 15 Jun 2018Obviously, the notions of support, maximality, and closedness, which have been introduced here for rankings, can be used for orders in exactly the same way.

Preference rules for label ranking: Mining patterns in multi-target relations

“…In this cases, we can use partial orders . We can represent partial orders with subrankings [5] or incomplete rankings [31] . For example, the partial order λ 1 λ 2 λ 4 can be represented as π = ( 1 , 2 , 0 , 3 ) , where 0 represents λ 1 , λ 2 , λ 4 ⊥ λ 3 .…”

“…Due to its intuitive representation, Association Rules [4] have become very popular in data mining and machine learning tasks (e.g. mining rankings [5] , classification [6] or even Label Ranking [7,8] ). In [7] , association rules were adapted for the prediction of rankings, which are referred to as Label Ranking Association Rules (LRAR).…”

“…Cases like this have motivated the appearance of new approaches, e.g. to mine ranking data [5] , where association rules are used to find patterns within rankings.…”

“…We propose a method which combines the two approaches mentioned above [5,7] , to extract interesting information from datasets even when the number of different rankings is very high. We define Pairwise Association Rules (PAR) as association rules with one or more pairwise comparisons in the consequent.…”

Learning interesting attributes for automated data categorization

“…The survey by Geng and Hamilton [13] provides a detailed discussion on subjective and objective measures used to capture 'interestingness' of data for association or classification rule mining. Henzgen and Hüllermeier [15] present an analogy of the itemset-mining measures support and interest applied to mining subrankings. Different context-specific diversity measures are proposed in the area of Web search [1,10,24], entity summarization [31], and recommender systems [27].…”