Chapter 9-10-complex data mining + application and development trend of data mining (9/10) + (10/10)

Spatial Data

Multimedia Data

For example, image data

Description-based retrieval system: keywords, titles, dimensions, etc.

Content-based retrieval system: color composition, texture, shape, object and wavelet transformation.

Time series data and sequence data

Trend Analysis

Long-term changes (long-term trends)

Cyclical changes (periodic changes, if any)

Seasonal changes

Irregular changes

Text Database Mining

Latent Semantic Indexing

The potential semantic index is used to reduce the size of the Word Frequency matrix. The core technology is Singular Value Decomposition. The procedure is as follows:

1. Create a word frequency matrix, frequency_matrix.

2. Calculate the Singular Value Decomposition of frequency_matrix by dividing the matrix into three small matrices U, S, and V. Where U and V are orthogonal matrices, and s is the diagonal matrix of Singular Values. The size of matrix S is k × K.

3 for each document D, replace the original vector with the new vector of the word eliminated in SVD.

4. Save the set of all vectors and use advanced multidimensional indexing technology to create indexes for them.