[week 11] Muddiest Points

This week the topic was "Web Search", and these are my two cents:

1. When did the search engines (before Google) started to consider the link structure of the web as an important issue to incorporate in their algorithms? I know HITS algorithm performs links analysis, but when did commercial search engines started to using it seriously? I want to find out whether this was the drop who balanced the scale in favor of Google.

2. In the model of the Web as a core, incoming and outgoing links... how is the 22% percent of disconnected pages calculated? if they are disconnected, how to be sure that is less or more?

3. How a search engine decides where to star crawling? Which are the most common heuristics to make this decision?

[week 10] Spring Break

No comments abot reasdings this week. I have been working on my final project, which I am developing with my classmate Parot Ratnapinda. We are using TREC Genomics dataset, Lucene project to index and Carrot 2 to clusterize and visualiza results.

[week 9] Reading Notes

This week the lecture is about Information Visualization. This is probably the area which most books get outdated in little time. Anyway, was very interesting to review the different kinds of data types and tasks of the taxonomy to identify visualization that is shown in the secondary reading of this week. A couple of questions:

1. I know a system called SpaceTime 3D, an application which gives enriched visualizations of the results of queries done in google and other Sites. Despite the name, I am not sure if this is a real 3D World data type or a 2D Map, or maybe a combination of both.

2. Does HayStaks ( a plugin for firefox to store and reutilize queries in an organization) corresponds to the challenge category of "Collaborate with others"?

[week 8] Midterm

Lots of muddiest points for the midterm... I will post something when correction is available.

[week 7] Muddiest Points

As the last post, this one is about Relevance Feedback. Firstly, I would like to know more details about the Wilcoxon signed-rank test. I know, I know... I can google it by myself.

Another question is that I guess that relevance feedback seems to be too attached to one specific user needs. If the algorithm updates weights based on user feedback, how can we be sure that if another user makes the same or a similar query has the same information need? Even if they have the same information need, How can we be sure that they will judge the results likely relevant? Even more, how is it possible that a model with two assumptions so weak (users like to provide feedback, we can obtain reliable feedback from users) can be successful?

[week 7] Reading Notes

Relevance Feedback

• In the section 9.1, the authors wrote the following: “RF can also be effective in tracking a user’s evolving information need”. I am not so sure of this, because if the user understands better her information need, will probably reformulate her query instead of waiting for the system to recalculate a result set.

Foundations of the Rocchio Algorithm

• I don’t understand how the formula (9.2)

is derived from the formula 6.10 of vector similarity.

The Rocchio Algorithm

• The book states that Relevance Feedback has shown to be useful for increasing recall in situations where it is important, and one of the reasons is because the technique expands the query. Is this always true? When I expand the query and I do a Boolean match of the query terms with AND, adding more terms can increase precision but not necessarily recall. Now, I know that in this case this is a VSM and not a Boolean one, but it is always true that adding terms to query the recall is increased?

• Besides, I don’t understand clearly the IDE DEC-HI concept.

[week 6] Muddiest Points

This week's lecture was about Evaluation of IR Systems. In addition to the weekly book reading, we read two papers:

• Karen Sparck Jones, What's the value of TREC: is there a gap to jump or a chasm to bridge? ACM SIGIR Forum, Volume 40 Issue 1 June 2006
• Kalervo Järvelin, Jaana Kekäläinen. Cumulated gain-based evaluation of IR techniques ACM Transactions on Information Systems (TOIS) Volume 20 , Issue 4 (October 2002) Pages: 422 – 446

TREC
Through the IIR book chapter and the first paper, I was introduced to TREC (Text REtrieval Conference) :
1. I was I little surprised when reading the paper about TREC with the affirmation of the author: <<...this progress... has not so far enabled the research community it represents to say:'if your retrieval case is like this, do this' as oposed to 'well, with tunning, this sort of thing could serve yo alright'. >> Despite Google's success, is this still the feeling about the IR community?
2. Which are the sources used to generate content in the Biological and Law tracks of TREC?
3. How do they (TREC consortium) decide to stop a track or to create a new one?

Interpolation on P-R graphs
About the Precision-Recall graphs, I wonder if the Interpolation process may produce, in some cases, a false interpretation of the performance of an IR system.