Hot mobile & Geo-location platform

"The idea that we can have a query free, geographically relevant search is really exciting." from Yahoo FireEagle: Geolocation made simple.

Ever since i bought my iPod touch, i realized that the next BIG thing is happening on mobile. I am usually not geek enough to be an earlier adopter of new technology, so when I am getting more and more stuff done on my beloved small gadget instead of the computer, it is the singal that it's about the time to GO Mobile.

Mobile computing, mobile search, mobile advertising, etc, yes, everything mobile related would get extremely hot in the next few years. Among all of these, geo-location platform might be one of the most important driving force as this is what most of mobile killer application based on.

From the research perspective, what interests me most is probably how people gonna share their geo-location data without causing too much trouble on personal privacy. It is right now the hot topic in academia. The most recent one would be this SIGMOD'08 paper

Private Queries in Location Based Services: Anonymizers are not Necessary

• Gabriel Ghinita, Panos Kalnis, Ali Khoshgozaran, Cyrus Shahabi, Kian-Lee Tan

I plan to attend their session in the next week, and will come back to this point later.

Quantitative measures of product features from the qualitative product reviews

Recently i came across an interesting kdd paper on deriving qualitative measures of product feature based on qualitative customer reviews.

Show me the Money! Deriving the Pricing Power of Product Features by Mining Consumer Reviews. KDD 07.

The focus of the paper is not on extracting the product features and the related phases, but to develop a model to evaluate the weight and strength of these features and related phases. The novelty of this paper is that they model the product demand as a function of the product features and related reviews ( by adapting some existing econometrics), and the quantitative measures for the features and reviews are thus can be estimated in relation to the product demand using the training set.

Here are some interesting observation from their experiments:
1. Products that get bad reviews, tend to disappear from the market quickly, and do not get many further negative evaluations. In general, the reviews that appear on Amazon are positive, especially for products with large number of posted reviews."

2.Most of the positive evaluations contain superlatives, and a mere "decent" or "fine" is actually interpreted by the buyers as a lukewarm, slightly negative evaluation.

3. The evaluations "best camera," "excellent camera," "amazing camera", "perfect camera," and so on, have a negative effect on demand as the customers discount such reviews ( if without details provided), and potentially treat them as bogus.

Migrated from Bloglines to Google

I really liked many of the features in Bloglines; but at some point, it seems to stop evolving. And then it comes the real trigger point: it is not supported well on my new IPod iTouch, which i fell in love recently, sigh......

For the previous posts back to April, 2008. Please refers to the following URL
http://www.bloglines.com/blog/Arber