Mining multimedia data

icde, 2000

Despite the overwhelming amounts of multimedia data recently generated and the significance of such data, very few people have systematically investigated multimedia data mining. With our previous studies on content-based retrieval of visual artifacts, we study in this paper the methods for mining content-based associations with recurrent items and with spatial relationships from large visual data repositories. A progressive resolution refinement approach is proposed in which frequent item-sets at rough resolution levels are mined, and progressively, finer resolutions are mined only on the candidate frequent item-sets derived from mining rough resolution levels. Such a multi-resolution mining strategy substantially reduces the overall data mining cost without loss of the quality and completeness of the results.

Knowledge and Information Systems, 2006

Generating captions or annotations automatically for still images is a challenging task. Traditionally, techniques involving higher-level (semantic) object detection and complex feature extraction have been employed for scene understanding. On the basis of this understanding, corresponding text descriptions are generated for a given image. In this paper, we pose the auto-annotation problem as that of multi-relational association rule mining where the relations exist between image-based features, and textual annotations. The central idea is to combine low-level image features such as color, orientation, intensity, etc. and corresponding text annotations to generate association rules across multiple tables using multi-relational association mining. Subsequently, we use these association rules to auto-annotate test images. In this paper we also present a multi-relational extension to the FP-tree algorithm to accomplish the association rule mining task effectively. The motivation for using multi-relational association rule mining for multimedia data mining is to exhibit the potential accorded by multiple descriptions for the same image (such as multiple people labeling the same image differently). Moreover, multi-relational association rule mining can also benefit the auto-annotation process by pruning the number of trivial associations that are generated if text and image features were combined in a single table through a join. In this paper, we discuss these issues and the results of our auto-annotation experiments on different test sets. Another contribution of this paper is highlighting a need to develop robust evaluation metrics for the image annotation task. We propose several applicable scoring techniques and then evaluate the performance of the different algorithms to study the utility of these techniques. A detailed analysis of the datasets used and the performance results are presented to conclude the paper.

2004

This paper presents algorithms for finding the meanings of the audio-visual video patterns obtained in the unsupervised discovery process. This problem is interesting in domains where neither perceptual patterns nor semantic concepts have simple structures. The patterns in the video are modeled with hierarchical hidden Markov models, with efficient algorithms to jointly learn the model parameters, the optimal model complexity, as well as the relevant feature subsets. The meanings are contained in words of the speech transcript of the video. The pattern-word association is obtained via co-occurrence analysis and machine translation models. Promising results are obtained on TRECVID news videos: video patterns that associate with distinct topics such as el-nino and politics are itentified; a temporal structure model compares favorably to a non-temporal clustering algorithm.

Semantic Mining Technologies for Multimedia Databases, 2009

ASSOCIATION RULES IMPLEMENTATION FOR AFFINITY ANALYSIS BETWEEN ELEMENTS COMPOSING MULTIMEDIA OBJECTS, 2019

The multimedia objects are a constantly growing resource in the world wide web, consequently it has generated as a necessity the design of methods and tools that allow to obtain new knowledge from the information analyzed. Association rules are a technique of Data Mining, whose purpose is to search for correlations between elements of a collection of data (data) as support for decision making from the identification and analysis of these correlations. Using algorithms such as: A priori, Frequent Parent Growth, QFP Algorithm, CBA, CMAR, CPAR, among others. On the other hand, multimedia applications today require the processing of unstructured data provided by multimedia objects, which are made up of text, images, audio and videos. For the storage, processing and management of multimedia objects, solutions have been generated that allow efficient search of data of interest to the end user, considering that the semantics of a multimedia object must be expressed by all the elements that composed of. In this article an analysis of the state of the art in relation to the implementation of the Association Rules in the processing of Multimedia objects is made, in addition the analysis of the consulted literature allows to generate questions about the possibility of generating a method of association rules for the analysis of these objects.

IJCSNS, 2010

Over the past decades, data mining has proved to be a successful approach for extracting hidden knowledge from huge collections of structured digital data stored in databases. From the inception, Data mining was done primarily on numerical set of data. Nowadays as large multimedia data sets such as audio, speech, text, web, image, video and combinations of several types are becoming increasingly available and are almost unstructured or semistructured data by nature, which makes it difficult for human beings to extract the information without powerful tools. This drives the need to develop data mining techniques that can work on all kinds of data such as documents, images, and signals. This paper explores on survey of the current state of multimedia data mining and knowledge discovery, data mining efforts aimed at multimedia data, current approaches and well known techniques for mining multimedia data.

2009 IEEE International Conference on Semantic Computing, 2009

Association rule mining (ARM) has been studied in the areas of content-based multimedia retrieval and semantic concept detection due to its high efficiency and accuracy. Two important processes in mining the association rules for classification are rule generation and rule selection. In this paper, a novel high-level feature detection framework using the ARM technique together with the correlations among the feature-value pairs is proposed. A new association rule mining (ARM) algorithm has been developed, where the N -featurevalue pairs are generated using a combined measure based on (1) the existence of the (N -1)-feature-value pairs (where N is larger than 1), (2) the correlation between different Nfeature-value pairs and the concept classes through Multiple Correspondence Analysis (MCA), and (3) the similarity representing the harmonic mean of the inter-similarity and intrasimilarity. The final association classification rules are selected by using the calculated harmonic mean of the similarity values. The proposed framework enables the automatic discovery and generation of the N -feature-value pair association rules from the 1-feature-value pairs for classification. Experimenting with 15 high-level features (concepts) and benchmark data sets from TRECVID, our proposed framework achieves promising performance and outperforms three other wellknown classifiers (Decision Trees, Support Vector Machine, and Neural Networks) which are commonly used for performance comparison in the TRECVID community.

2007 IEEE 11th International Conference on Computer Vision, 2007

We present a novel approach to automatically find spatial configurations of local features occurring frequently on instances of a given object class, and rarely on the background. The approach is based on computationally efficient data mining techniques and can find frequent configurations among tens of thousands of candidates within seconds. Based on the mined configurations we develop a method to select features which have high probability of lying on previously unseen instances of the object class. The technique is meant as an intermediate processing layer to filter the large amount of clutter features returned by lowlevel feature extraction, and hence to facilitate the tasks of higher-level processing stages such as object detection.

Multimedia data mining is a popular research domain which helps to extract interesting knowledge from multimedia data sets such as audio, video, images, graphics, speech, text and combination of several types of data sets. Normally, multimedia data are categorized into unstructured and semi-structured data. These data are stored in multimedia databases and multimedia mining is used to find useful information from large multimedia database system by using various multimedia techniques and powerful tools. This paper provides the basic concepts of multimedia mining and its essential characteristics. Multimedia mining architectures for structured and unstructured data, research issues in multimedia mining, data mining models used for multimedia mining and applications are also discussed in this paper. It helps the researchers to get the knowledge about how to do their research in the field of multimedia mining.

Data Mining and Knowledge Discovery: Theory, Tools, and Technology III, 2001