Apparently the creator of the Simpson, at the beginning of his career, sketched an Ad for Apple. This is from 1989. Imagine doing what they promise with the Macs they advertise at the end of the booklet. Also, “The Simpsons” were just barely spun off from being cartoon filler for the Tracy Ullman show. Matt Groening was still primarily known as the underground cartoonist of “Life In Hell“.
Human beings like to make things: they like to use their hands at least as much as their brains;…If we are to tap the increasing visuality and dynamics of computing in order to open new realms of abstraction, we should depend very much on these human traits to do so.
McCullough, 1996
Voting with an SMS is a very simple and yet powerful idea that enables lots of people to express themselves on current political decisions and issues. It is easy to implement and it is costless for the government. It was tested recently in Wales. I wish I could have something like that in Italy.
Voters will be asked to text “Wales Yes” or “Wales No” to 83010, votes will be monitored to ensure that no more than one vote from each mobile number is counted and each vote will cost 25p plus the standard network charge.
A. Kontosthathis, W. M. Pottenger, and B. D. Davison. Assessing the impact of sparsification on lsi performance. In Proceedings of the 2004 Grace Hopper Celebration of Women in Computing Conference, Chicago, IL, USA, Oct 6-9 2004. [url]
————————————–
This paper presents a technique for sparsification of LSI values in the Singular Value Decomposition. The algorithm implemented defines a positive and a negative threshold. Any element Tk in the Term by Dimension matrix that falls in the interval defined by the thresholds is set to zero. Results shown not significant losses in the precision and recall of the information retrieval.
ABSTRACT
We describe an approach to information retrieval using Latent Semantic Indexing (LSI) that directly manipulates the values in the Singular Value Decomposition (SVD) matrices. We convert the dense term by dimension matrix into a sparse matrix by removing a fixed percentage of the values. We present retrieval and runtime performance results, using seven collections, which show that using this technique to remove up 70% of the values in the term by dimension matrix results in similar or improved retrieval performance (as compared to LSI), while reducing memory requirements and query response time. Removal of 90% of the values results in significantly reduced memory requirements and dramatic improvements in query response time. Removal of 90% of the values degrades retrieval performance slightly for smaller collections, but improves retrieval performance by 60% on the large collection we tested.
Tags: clustering, information retrieval, Latent Semantic Analysis, Singular Value Decomposition
G. Pivovarov and S. Trunov. Eqrank: A self-consistent equivalence relation on graph vertexes. ACM SIGKDD Explorations, 5(2):185–190, 2003. [url]
——————–
This paper presents a new method for hierarchical clustering of directed graph. The authors tried to answer the following research question: do there exist nontrivial hierarchical graph clustering that would be uniquely determined by the graph structure?
They answer the question proposing a new algorithm, called EqRank that is tested to cluster scientific papers. Results seems to confirm the idea that graph based method can be used concurrently to canonical LSI methods, reducing drammatically the computational load required to achieve the clustering.
Tags: clustering, information retrieval
Here are a couple of pointers for the Part of Speech Tagger. The first one it is the one I am using for my project. It was developed by the University of Stuttgart and it has the advantage to support multiple languages as it is based on a probabilistic criteria.
http://www.ims.uni-stuttgart.de/projekte/corplex/TreeTagger/DecisionTreeTagger.html
The following, from the University of Edinburgh is just for the English language. The page has also some good pointers to some literature on the subject:
http://www.ltg.ed.ac.uk/software/pos/
Here is a general overview on the subject:
http://www.georgetown.edu/faculty/ballc/ling361/tagging_overview.html
Tags: tagging, text data mining
I just received an invitation to this public display of the research project of the Okude Laboratory. This exhibition is going to be part of the Open Research Forum, an annual academic meeting of Keio University.
The concept of “New Mode of Life” is central to the ideas developed by the Okude lab, new design of media giving happiness in our life.
Jesse James Garrett maintains a wonderful tutorial on how to describe visually interaction design.
Diagrams are an essential tool for communicating information architecture and interaction design in Web development teams. This document discusses the considerations in development of such diagrams, outlines a basic symbology for diagramming information architecture and interaction design concepts, and provides guidelines for the use of these elements.
Eric Turner’s quick reference guide is an easy one-page description of all the elements. And Scott Larson created this handy cheatsheet for quick reference to the various conditional elements.
Tags: information visualization, interaction design, flowcharts
G. Raimondon. The Traveling Salesman Problem. Technical Report June, ROSE, Ecole Politechnique Federale de Lausanne, Ecoublens, CH-1015, Switzerland, 2005.
————————-
This tech report contains a technical solution for the TSP (Travelling Salesman Problem). The solution of the problem requires an algorithm that is computationally heavy. This paper describe the implementation of the 2-opt solution that is able to find a local minima of the problem that reduces drammatically the number of the computations required.
Term Frequency-Inverse Document Frequency. A kind of DocumentVector. This scheme assigns a weight to each term (vocabulary word) in a given document. The weight increases proportional to the number of times the term occurs in the document, but is offset by a term which devalues terms common in the overall corpus.
One formula (apparently a simplification of (Salton and Buckley, ’88)) is the following. The weight of a term t in a document D is:
(# of occurrences of term t in this document D) * log((total # of documents)/(# of documents with mention of term t))
References:
Gerard Salton , Christopher Buckley, Term-weighting approaches in automatic text retrieval, Information Processing and Management: an International Journal, v.24 n.5, p.513-523, 1988
Copyright notice: the present content was taken from the following URL, the copyrights are reserved by the respective author/s.
