Interestingness measures for data mining

Luo, Geng; Hamilton, Howard J.

doi:10.1145/1132960.1132963

Cited by 942 publications

(704 citation statements)

References 28 publications

Supporting

Mentioning

682

Contrasting

Unclassified

Order By: Relevance

“…The obtained patterns are semantically richer nevertheless this type of extraction leads to the exploration of a huge search space with an important amount of patterns. In the future, we wish to adapt some interestingness measures [15,16] to these kinds of patterns to 1) filter the patterns according to experts' needs and 2) push it in the pattern extraction process. We aim at improving the extraction time by reducing the search space, and also provide experts with the minimal and most interesting set of spatio-temporal and related patterns.…”

Section: Resultsmentioning

confidence: 99%

Including Spatial Relations and Scales within Sequential Pattern Extraction

et al. 2012

View full text Add to dashboard Cite

Abstract. Georeferenced databases contain a huge volume of temporal and spatial data. They are notably used in environmental analysis. Several works address the problem of mining those data, but none are able to take into account the richness of the data and especially their spatial and temporal dimensions. In this paper, we focus on the extraction of a new kind of spatio-temporal pattern, considering the relationship between spatial objects and geographical scales. We propose an algorithm, STR PrefixGrowth, which can be applied to a huge amount of data. The proposed method is evaluated on hydrological data collected on the Saône watershed during the last 19 years. Our experiments emphasize the contribution of our approach toward the existing methods.

show abstract

Section: Resultsmentioning

confidence: 99%

Including Spatial Relations and Scales within Sequential Pattern Extraction

et al. 2012

View full text Add to dashboard Cite

show abstract

“…Besides, we need measures which assign high weights for relations among specific terms and low weights for relations among generic terms. We can employ similarity (also called quality, interestingness or association) measures to compute relations among terms [7,27]. Different measures calculate the similarity between terms t i and t j (Ω(t i , t j )) considering the information contained in the contingency matrix presented in Table 1.…”

Section: Term Network Generationmentioning

confidence: 99%

“…We selected measures which comply with different characteristics and properties [7,27]. The selected measures were: Support, Yule's Q, Mutual Information, Kappa, and Piatetsky-Shapiro.…”

Section: Term Network Generationmentioning

confidence: 99%

Term Network Approach for Transductive Classification

Rossi¹,

Rezende²,

Lopes³

2015

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Abstract. Transductive classification is a useful way to classify texts when just few labeled examples are available. Transductive classification algorithms rely on term frequency to directly classify texts represented in vector space model or to build networks and perform label propagation. Related terms tend to belong to the same class and this information can be used to assign relevance scores of terms for classes and consequently the labels of documents. In this paper we propose the use of term networks to model term relations and perform transductive classification. In order to do so, we propose (i) different ways to generate term networks, (ii) how to assign initial relevance scores for terms, (iii) how to propagate the relevance scores among terms, and (iv) how to use the relevance scores of terms in order to classify documents. We demonstrate that transductive classification based on term networks can surpass the accuracies obtained by transductive classification considering texts represented in other types of networks or vector space model, or even the accuracies obtained by inductive classification. We also demonstrated that we can decrease the size of term networks through feature selection while keeping classification accuracy and decreasing computational complexity.

show abstract

“…Because they are general and important concepts, beauty and interestingness have previously been explored in diverse contexts including philosophy (Neill and Ridley, 1995), reinforcement learning (Schmidhuber, 2009), and even data mining (Geng and Hamilton, 2006).…”

Section: Impressiveness and Interestingnessmentioning

confidence: 99%

“…Though they overlap in some ways, a key difference between interestingness and impressiveness is that interestingness is often tied to time-dependence or novelty (Geng and Hamilton, 2006). That is, an object that is initially found interesting may become less interesting over time due to habituation.…”

Section: Impressiveness and Interestingnessmentioning

confidence: 99%

Beyond Open-endedness: Quantifying Impressiveness

Lehman¹,

Stanley²

2012

Artificial Life 13

View full text Add to dashboard Cite

This paper seeks to illuminate and quantify a feature of natural evolution that correlates to our sense of its intuitive greatness: Natural evolution evolves impressive artifacts. Within artificial life, abstractions aiming to capture what makes natural evolution so powerful often focus on the idea of openendedness, which relates to boundless diversity, complexity, or adaptation. However, creative systems that have passed tests of open-endedness raise the possibility that openendedness does not always correlate to impressiveness in artificial life simulations. In other words, while natural evolution is both open-ended and demonstrates a drive towards evolving impressive artifacts, it may be a mistake to assume the two properties are always linked. Thus to begin to investigate impressiveness independently in artificial systems, a novel definition is proposed: Impressive artifacts readily exhibit significant design effort. That is, the difficulty of creating them is easy to recognize. Two heuristics, rarity and re-creation effort, are derived from this definition and applied to the products of an open-ended image evolution system. An important result is that that the heuristics intuitively separate different reward schemes and provide evidence for why each evolved picture is or is not impressive. The conclusion is that impressiveness may help to distinguish open-ended systems and their products, and potentially untangles an aspect of natural evolution's mystique that is masked by its co-occurrence with open-endedness.

show abstract

Interestingness measures for data mining

Cited by 942 publications

References 28 publications

Including Spatial Relations and Scales within Sequential Pattern Extraction

Including Spatial Relations and Scales within Sequential Pattern Extraction

Term Network Approach for Transductive Classification

Beyond Open-endedness: Quantifying Impressiveness

Contact Info

Product

Resources

About