(I) C.D. Manning, P. Raghavan, P. Raghavan Introduction to Information Retrieval, Cambridge University Press - 2008
(I) A. Rajaraman, J. D. Ullman, Mining of Massive Datasets, 2011
(L) I. Witten, A. Moffat, T.C. Bell Managing Gigabytes, Van Nostrand Reinhold – 1999
(L) D. Doermann, K. Tombre (Eds.) Handbook of Document Image Processing and Recognition, 2014 (L)
(L) : Book available in the Engineering library
(I) : Book available in Internet (authors' version)
The course first aims at introducing the main Data Mining techniques that allow you to model large amounts of data and extract useful information.
Secondly, we consider the problems arising when extracting information and indexing both textual and non-textual documents. To this purpose we introduce the main models and algorithms in Information Retrieval and describe the techniques for information extraction from digital born and digitized documents that are represented in the form of images.
It is essential to know topics typically though in the Data Bases and Algorithms and Data Structures classes. Some knowledge of Artificial Intelligence can be useful.
Classes, homework and project.
Oral exams are usually made after completion of the assigned project.
Type of Assessment
Study and presentation of one research paper to the class (15%). Group project (2 people, 65%). Oral on a sub-set of the topics (20%).
Alternatively it is possible to have an oral on all the topics and a smaller project.
Datawarehouse. Hardware. Disk Organization. Access times
Distributed file system and the new software stack
Map Reduce, Word count, Matrix-Vector and Matrix Multiplication with Map Reduce
The market-basket model . Association rules. Implementation details. Algorithms for computing frequent item-sets and Association Rules.
Improving Apriori: Hash-based filtering. Bloom filters. PCY algorithm, Random sampling, SON algorithm, Apriori with MapReduce-
Finding similar items. Curse of dimensionality. Distance measures.
Document similarity, shingling, min-hashing
Locality sensitive hashing (LSH)
Families of hash functions. LSH for cosine distance. LSH for Euclidean distance.
Curse of dimensionality. Distance measures.
Clustering, Hierarchical clustering, k-means clustering. SOM clustering
BFR algorithm, CURE algorithm. Dimensionality reduction
Document Image Analysis and Recognition
Layout analysis : RLSA, Docstrum, Area Voronoi diagram, XY tree, MXY tree, Reading order detection, classification in layout analysis, page classification/retrieval.
Layout analysis : XY tree, MXY tree, Reading order detection, classification in layout analysis, page classification/retrieval. OCR.
Artificial Neural Networks. Perceptron, Backpropagation
Convolutional neural networks
Document Image Retrieval
Introduction to Information Retrieval. Boolean Retrieval
Term vocabulary and postings lists, Inverted files
Vector Space Model
Tokenization, stop-word removal, stemming
Processing boolean queries
Computing Scores in complete search system - Efficient scoring and ranking, Components of an information retrieval system. Vector space scoring and query operator interaction.
Performance Evaluation in IR systems