Web Science/Part2: Emerging Web Properties/Generative Models for the Web/Evaluating a generative model

27

Web Science/Part2: Emerging Web Properties

Web Science/Part2: Emerging Web Properties/MoocIndex

Web Science/Part2: Emerging Web Properties/Generative Models for the Web/Evaluating a generative model

• edit MOOC index

Evaluating a generative model

Learning goals

Video

Script

Quiz

Further reading

Discussion

Learning goals

1. See that it makes sense to compare statistics
2. Understand that comparing statistics is not a well defined task
3. Be aware of the fact that very different models could lead to the same statistics

Video

Script

The slide deck can be found at File:Evaluating-a-generative-model.pdf

Quiz

What is true if the statistics of a generative Model match the statistics of the descriptive model?

	The generated data can still be very different from the observed data.
	The generated model is a perfect match of the observed data
	There could be other statistics which we haven't looked at that might not match.
	The model parameters of the generative model could give some reason why we can observe something
	One should try to decrease the number of model parameters

Further reading

1. tba

Discussion

Navigation

1. How big is the World Wide Web
   1. Problems with the question about the size of the Web
   2. 3 ways to study the Web
   3. A simplistic descriptive model
   4. An unrealistic, simplistic generative model
   5. Summary, further reading, homework
2. Simple statistical descriptive Models for the Web
   1. Counting Words And Documents
   2. Typical length of a document
   3. How to formulate a research hypothesis
   4. Number of words needed to understand most of Wikipedia
   5. Linguists way of checking simplicity of text
3. Advanced statistical descriptive models for the Web
   1. The Zipf law for text
   2. Visually straight lines on log log plots
   3. Fitting a curve on a log log plot
   4. Zipf law powerlaw or pareto law.webm
4. Modelling Similarity of Text
   1. Similarity Measures and their Applications
   2. Jaccard Similarity for Sets
   3. Cosine Similarity For Vectorspaces
   4. Probabilistic Similarity Measures Kullback Leibler Divergence
   5. Comparing Results of Similarity Merasures
5. Generative Models for the Web
   1. Introduction to generative modelling.webm
   2. Sampling from a probability distribution
   3. Evaluating a generative model
   4. Pittfalls when increasing the number of model parameters
6. Modeling the Web as a graph
   1. Reviewing terms from graph theory
   2. The standard web graph model
   3. Descriptive statistics of the web graph
   4. Topology of the web graph
   5. Modelling-graphs-with-linear-algebra

Sampling from a probability distribution

Pittfalls when increasing the number of model parameters