Deep Web Mining and Learning for Advanced Local

Deep Web Mining and Learning for Advanced Local Search CS 8803 Advisor Prof Liu Yu Liu, Dan Hou Zhigang Hua, Xin Sun Yanbing Yu

Competitors n n n Yahoo! Local Yelp City. Search Google Local Yellow Page How to beat them?

Research Background n n n Deep Web Crawling Sentimental Learning Sentimental Ranking Model Geo-credit Ranking Model Social Network for Businesses

Show Time! Local Biz Space

Architecture Query-based Crawler Sentimental Learner Super Local-Search HTML Parser Apache Server JDBC Database

Tools Open source social network platform Elgg, Open. Social n LAMP Server Linux+Apache+Mysql+PHP n Google Map API, eg, Geocode, n

Crawling Dynamic Pages

Parsing Dynamic Pages

Database Design

Sentimental Learning

Sentimental Learning Can we use ONE score to show good/ bad the store is?

Sentimental Learning n Objective • To identify positive and negative opinions of a store n Dataset • Reviews represented by bag-of-terms • Normalized TF-IDF feature (normalized) n Two ways of sentiment representation • Simply average the scores n but “what you think good might be bad for me” • Manual labeling n n n 1 to 5 (“least satisfied” to “most satisfied”) consensus based time-accuracy tradeoff

Dimension Reduction n High dimensionality • 6857 tokens n n n Memory limitation Possibly under-fitting Dimension Reduction • PCA (Principle Component Analysis) n n n an orthogonal linear transformation transforms the data to a new coordinate system retains the characteristics of the data set that contribute most to its variance • Get the most important features without losing generality

Principle Component Analysis n n n Original Dimension: 6857 Covariance Reserved: 95% Different Granularity • Manual Labeling: Downsampling Rate Result Dimension 40% 231 20% 4% 135 41 20% 211 4% 37 • Score Averaging: Downsampling Rate Result Dimension 40% 344

Sentimental learning n Features used for sentimental learning: • Vector Space Model (reviews/comments) n Some keywords related to sentiments: • Positive: good, happy, wonderful, excellent, awesome, great, ok, nice, etc • Negative: bad, sad, ugly, outdated, shabby, stupid, wrong, awful, etc n Most words unrelated to sentiments: • e. g. buy, take, go, i. Pod, apple, comment, etc… • Causing noise for sentimental learning!!

What we do? n How to learn sentiments from a large set of features with lots of noise? n n Vector Space Model: MXN (Entity-Term, e. g. 6, 000 X 20, 000) Dimensionality reduction (PCA) Using supervised learning for sentimental learning Human labeling vs. Average rating • An online entity always includes many reviews with each review containing a rating n Average Rating is an alternative labeling for the entity • Manual labeling: n n 1 (least satisfactory) – 5 (most satisfactory) Three persons do labeling, most-vote-adopted

Manual labeling vs. Average rating n Machine learning • Around 300 entities from local search, 6800 features after stop words removing and stemming • Using different SVM kernels • Avoiding overfit n Leave-one-out estimation • Nonlinearity of features • Polynomial kernel achieves best performance n Manual labeling • Training more precise • Labeling more consistent n Rate averaging • Training less precise • Rating more random • E. g. average(5, 5, 1) = 3

What we learned? n Dimensionality reduction is necessary • Term Vector Space Model (VSM) is huge in nature n Human labeling is necessary • Sentimental learning involved subjective judge instead of objective judge. • Human rating is very random because it is not consistent across different people • More labeling data is needed n Other methods to be used: • Unsupervised learning (clustering) n Gaussian Mixture Model (an alternative to learn sentiments, while it is difficult to know the # of hidden sentiments)

How to use learned sentiments? n Sentimental learning can be used to improve ranking of local search • Because sentimental value represents an important metrics to evaluate the rank of an entity • Local search is influenced by the sentiment n Sentimental ranking model (SRM): • Senti. Rank = a*Content. Sim + (1 -a)*Senti. Value n Empirically setting the parameter as “ 0. 5”. • Similar to Page. Rank n Page. Rank = b*Content. Sim + (1 -b)*Page. Importance

Geocoding n n Geocoding of Addresses For example , the geo-center of store AA National Auto Parts Is located at 3410 Washington St, Phoenix, AZ, 85009 Using Geocode, we can get the exact latitude and longtitude (33. 447708, -112. 13246) Haversine Formula of Great-circle distance: Distance between two pairs of coordinates on sphere = (3959 * acos( radians(33. 448) ) * cos( radians( lat ) ) * cos( radians( lng ) - radians(-122) ) + sin( radians(-112. 132) ) * sin(radians( lat ) )

Geo-Sentimental Ranking Model (GSRM) n Three Measurements 3. Content Similarity Sentimental Value Geo-distance n GSRM Ranking model 1. 2. -- term-frequency -- sentimental learning -- Google Map API

Example

n n Thank You ! QA time