Research Problems in Data Mining

1
Research Problems in Data Mining

• Jiawei Han
• Database Systems Research Lab
• Department of Computer Science
• University of Illinois at Urbana-Champaign,
  U.S.A.
• http//www.cs.uiuc.edu/hanj

2
Several Research Issues in Data Mining

• Web mining and text mining
• Biomedical/DNA data mining
• On-line, real-time, stream data mining
• Cube exploration iceberg, cube-gradient, trends,
  etc.
• Mining max/closed long and error-tolerant
  frequent and sequential patterns
• Intrusion detection and anomaly mining
• Invisible data mining

3
Challenges in Web Mining

• Web A huge, widely-distributed, highly
  heterogeneous, semi-structured, interconnected,
  evolving, hypertext/hypermedia information
  repository.
• Problems
• the abundance problem
• limited coverage of the Web (hidden Web sources)
• limited query interface keyword-oriented search
• limited customization to individual users
• DBMS, DBers, and data miners will play an
  increasingly important role in the new generation
  of Internet

4
Web Mining Lots Can Be Done!

• A taxonomy of Web mining
• Web content mining
• Web usage mining
• Some interesting problems on Web mining
• Mining what Web search engine finds
• Weblog mining (usage, access, and evolution)
• Identification of authoritative Web pages
• Web document classification
• Warehousing a Meta-Web Web yellow page service
• Intelligent query answering in Web search

5
Mine What Web Search Engine Finds

• Current Web search engines convenient source for
  mining
• keyword-based, return too many answers, low
  quality answers, still missing a lot, not
  customized, etc.
• Data mining will help
• coverage enlarge and then shrink, using
  synonyms and conceptual hierarchies
• better search primitives user preferences/hints
• linkage analysis authoritative pages and
  clusters
• Web-based languages XML WebSQL WebML
• customization home page Weblog user profiles

6
Web Log Mining

• Weblog provides rich information about Web
  dynamics
• Multidimensional Weblog analysis
• disclose potential customers, users, markets,
  etc.
• Web accessing association/sequential pattern
  analysis
• Web cashing, prefetching, swapping
• Web linkage adjustment
• Trend analysis
• Dynamics of the Web what has been changing?
• Customized to individual users
• Need additional information in order to discover
  truly useful patterns

7
Discovery of Authoritative Pages in WWW

• Page-rank method ( Brin and Page, 1998)
• Rank the "importance" of Web pages, based on a
  model of a "random browser."
• Hub/authority method (Kleinberg, 1998)
• Prominent authorities often do not endorse one
  another directly on the Web.
• Hub pages have a large number of links to many
  relevant authorities.
• Thus hubs and authorities exhibit a mutually
  reinforcing relationship
• Both the page-rank and hub/authority
  methodologies have been shown to provide
  qualitatively good search results for broad query
  topics on the WWW, e.g., Google.

8
Web Document Classification

• Automatic classification of Web pages vs. human
  classification (e.g., Yahoo)
• Training set
• Existing, typical good classification sites,
  e.g., Yahoo!, CS term hierarchies
• Classification methods
• Typical method Naïve Bayesian, decision trees,
  etc.
• Key-word based classification is different from
  multi-dimensional classification
• Association- or clustering- based classification
  is often more effective
• Multi-level classification is important

9
Warehousing a High-Level Web An MLDB Approach

• ML-Web A structure which summarizes the
  contents, structure, linkage, and access of the
  Web and which evolves with the Web
• Layer0 the Web itself
• Layer1 the lowest layer of the ML-Web
• An entry a Web page summary, including class,
  time, URL, contents, keywords, popularity, rank,
  links, etc.
• Layer2 and up summary/classification/clustering
  in various ways and distributed for various
  applications
• ML-Web can be warehoused and incrementally
  updated
• Querying and mining can be performed on or
  assisted by ML-Web (a ML digital library
  catalogue, yellow page).

10
A Multiple Layered Meta-Web Architecture
More Generalized Descriptions
Layern
...
Generalized Descriptions
Layer1
Layer0
11
Construction of Multi-Layer Meta-Web

• XML facilitates structured and meta-information
  extraction
• Hidden Web DB schema extraction other meta
  info
• Automatic classification of Web documents
• based on Yahoo!, etc. as training set
  keyword-based correlation/classification analysis
  (IR/AI assistance)
• Automatic ranking of important Web pages
• authoritative site recognition and clustering Web
  pages
• Generalization-based multi-layer ML-Web
  construction
• With the assistance of clustering and
  classification analysis

12
Use of Multi-Layer Meta Web

• Benefits of Multi-Layer Meta-Web
• Multi-dimensional Web info summary analysis
• Approximate and intelligent query answering
• Web high-level query answering (WebSQL, WebML)
• Web content and structure mining
• Observing the dynamics/evolution of the Web
• Is it realistic to construct such a meta-Web?
• Benefits even if it is partially constructed
• Benefits may justify the cost of tool
  development, standardization and partial
  restructuring

13
Intelligent Web Query Answering

• What is intelligent query answering?
• Smart alternative answers, summary information,
  etc.
• Based on users profiles or history
• Web query needs more intelligent query answering
  mechanism
• How to develop it?
• Data warehouse and Web Yellow Page service will
  help
• Data mining will help too!

14
Biomedical Data Mining and DNA Analysis

• DNA sequences 4 basic building blocks
  (nucleotides) adenine (A), cytosine (C), guanine
  (G), and thymine (T).
• Gene a sequence of hundreds of individual
  nucleotides arranged in a particular order
• Humans have around 50,000 genes
• Tremendous number of ways that the nucleotides
  can be ordered and sequenced to form distinct
  genes
• Semantic integration of heterogeneous,
  distributed genome databases
• Current highly distributed, uncontrolled
  generation and use of a wide variety of DNA data
• Data cleaning and data integration methods
  developed in data mining will help

15
Discovery and Comparison of DNA Sequences

• Finding tandem repeats
• Fault-tolerant sequential patterns (Is Blast
  enough?)
• Similarity search and comparison among DNA
  sequences
• Compare the frequently occurring patterns of each
  class (e.g., diseased and healthy)
• Identify gene sequence patterns that play roles
  in various diseases

16
Association and Path Analysis in Bio-Medical and
DNA Data Mining

• Association analysis identification of
  co-occurring gene sequences
• Most diseases are not triggered by a single gene
  but by a combination of genes acting together
• Association analysis may help determine the kinds
  of genes that are likely to co-occur together in
  target samples
• Path analysis linking genes to different disease
  development stages
• Different genes may become active at different
  stages of the disease
• Develop pharmaceutical interventions that target
  the different stages separately
• Visualization tools and genetic data analysis

17
Stream Data and Applications

• Prevalence of Data Streams
• Network flow analysis and management
• Telephone call details and fraud detection
• On-line sensor monitoring
• Characteristics of Streams
• Massive and continuous amounts of data
• O(100 GB) per day
• Storage constraints
• small space (e.g., main memory or cache)
• Online continuous queries
• i.e., agg stream -gt stream

18
On-Line Mining of Stream Data

• Single-pass aggregates
• Basic, simple (exact) aggregates
• trivial inherently single-pass
• Approx fancy aggregates (e.g., median)
• Discovering correlations, associations, models,
  cause-and-effect relationships between patterns
• Discovering trends, clusters, changes, and
  outliers in data flowvisual mining will help as
  well
• Data stream warehousingsaving regularities?
  constraint- or goal- directed stream mining?

19
Multidimensional Data and Data Cubes

• Sales volume as a function of product, month, and
  region

Dimensions Product, Location, Time Hierarchical
summarization paths
Region
Industry Region Year Category
Country Quarter Product City Month
Week Office Day
Product
Month
20
Mining and Explorative Analysis of Data Cubes

• Efficient computation of data or iceberg cubes
• Discovery-driven data cube analysis
• Cube-gradient analysis
• What are the changes of the average house value
  in Sillicon Valley in 2001 comparing with 2000?
• Under what conditions the average house value
  increases 10 per year in Chicago area in 1990s?

21
What More can Be Done at Mining Data Cubes?

• Trend analysis in data cubes?
• What kind of companies have similar asset
  increase trends like Microsoft?
• Cluster customers based on their similar shopping
  behavior with regard to the change of time
• Model-based class comparison
• What are the critical features that distinguish
  winners and losers?
• Association and correlation analysis in data
  cubes
• If companys average profit is high, what other
  features will go with it?

22
Further Development of Frequent and Sequential
Pattern Analysis

• Efficient frequent pattern mining methods
• Association Apriori (94), FP-growth (00)
• Sequential pattern GSP(96), PrefixSpan (01)
• Mining max patterns, closed patterns,
  approximately closed, top-n frequent patterns
• Error-tolerant frequent and sequential patterns
  (e.g., DNA sequences)
• Constraint-based mining of frequent and
  sequential patterns

23
Intrusion Detection and Anomaly Mining

• Fighting against crimes and terrorists
• Linking and mining dynamic and huge amounts of
  data
• Sifting irregularities from regular ones Mining
  regularities as base for comparison and find
  outliers
• Classification normal vs. alarming classes and
  models
• Clustering and outlier analysis
• Human-instructed conditions and condition-guided
  classification, clustering, and outlier analysis
• Information visualization and stream data analysis

24
Invisible Data Mining

• Embed mining functions into information services
• Web search engine (link analysis, authoritative
  pages, user profiles)adaptive web sites, etc.
• Improvement of query processing history data
• Making service smart and efficient
• Benefits from/to data mining research
• Data mining research has produced many scalable,
  efficient, novel mining solutions
• Applications feed new challenge problems to
  research

25
Conclusions

• Data mining A young and promising discipline
• A confluences of multiple disciplines database,
  data warehouse, machine learning, statistics,
  high performance computing, Web technology, etc.
• Great progress in the last decade
• Lots of research issues, and a few identified
  here
• Web mining and text mining
• Biomedical/DNA data mining
• On-line, real-time, stream data mining
• Cube exploration iceberg, cube-gradient, trends,
  etc.
• Mining max/closed long and error-tolerant
  frequent and sequential patterns
• Intrusion detection and anomaly mining
• Invisible data mining

26
http//www.cs.uiuc.edu/hanj

• Thank you !!!

27
Selected Publications (2001)

• A. K. H. Tung, J. Hou, and J. Han, "Spatial
  Clustering in the Presence of Obstacles", Proc.
  2001 Int. Conf. on Data Engineering (ICDE'01),
  Heidelberg, Germany, April 2001.
• J. Pei, J. Han, H. Pinto, Q. Chen, U. Dayal, and
  M.-C. Hsu, "PrefixSpan Mining Sequential
  Patterns Efficiently by Prefix-Projected Pattern
  Growth", Proc. 2001 Int. Conf. on Data
  Engineering (ICDE'01), Heidelberg, Germany, April
  2001.
• J. Pei, J. Han, and L. V. S. Lakshmanan, "Mining
  Frequent Itemsets with Convertible Constraints",
  Proc. 2001 Int. Conf. on Data Engineering
  (ICDE'01), Heidelberg, Germany, April 2001.
• A. K. H. Tung, J. Han, L. V. S. Lakshmanan, and
  R. T. Ng, "Constraint-Based Clustering in Large Da
  tabases", Proc. 2001 Int. Conf. on Database
  Theory (ICDT'01), London, U.K., Jan. 2001.
• H. Miller and J. Han (eds.), Geographic Data
  Mining and Knowledge Discovery, Taylor and
  Francis, 2001.
• Y. Bedard, T. Merrett, and J. Han, "Fundamentals
  of Geospatial Data Warehousing for Geographic
  Knowledge Discovery", H. Miller and J. Han
  (eds.), Geographic Data Mining and Knowledge
  Discovery, Taylor and Francis, 2001.
• J. Han, M. Kamber, and A. K. H. Tung, "Spatial
  Clustering Methods in Data Mining A Survey", H.
  Miller and J. Han (eds.), Geographic Data Mining
  and Knowledge Discovery, Taylor and Francis,
  2001.
• H. Lu, L. Feng, and J. Han, "Beyond
  Intra-Transaction Association AnalysisMining
  Multi-Dimensional Inter-Transaction Association
  Rules", ACM Transactions on Information Systems,
  2001.

28
Selected Publications (2000)

• J. Han and M. Kamber, Data Mining Concepts and
  Techniques, Morgan Kaufmann, August 2000.
• K. Wang, Y. He and J. Han, "Mining Frequent
  Itemsets Using Support Constraints", Proc. 2000
  Int. Conf. on Very Large Data Bases (VLDB'00),
  Cairo, Egypt, Sept. 2000, pp. 43-52.
• E. D. Kim, J. M.W. Lam, and J. Han, "AIM
  Approximate Intelligent Matching for Time Series
  Data", Proc. 2000 Int. Conf. on Data Wareshouse
  and Knowledge Discovery (DaWaK'00), Greenwich,
  U.K., Sept. 2000.
• J. Han, J. Pei, B. Mortazavi-Asl, Q. Chen, U.
  Dayal, M.-C. Hsu, "FreeSpan Frequent
  Pattern-Projected Sequential Pattern Mining",
  submitted to 2000 Int. Conf. on Knowledge
  Discovery and Data Mining (KDD'00), Boston, MA,
  August 2000.
• J. Pei and J. Han "Can We Push More Constraints
  into Frequent Pattern Mining?", submitted to 2000
  Int. Conf. on Knowledge Discovery and Data Mining
  (KDD'00), Boston, MA, August 2000.
• J. Han, J. Pei, and Y. Yin, "Mining Frequent
  Patterns without Candidate Generation", Proc.
  2000 ACM-SIGMOD Int. Conf. on Management of Data
  (SIGMOD'00), Dallas, TX, May 2000.
• J. Pei, J. Han, and R. Mao, "CLOSET An Efficient
  Algorithm of Mining Frequent Closed Itemsets for
  Association Rules", submitted to 2000 ACM-SIGMOD
  Int. Workshop on Data Mining and Knowledge
  Discovery (DMKD'00), Dallas, TX, May 2000.
• D. Cheung, C. Hwang, A. Fu, and J. Han,
  "Efficient Rule-Based Attributed-Oriented
  Induction for Data Mining", Journal of
  Intelligent Information Systems, 15(2) 175-200,
  2000.
• N. Stefanovic, J. Han, and K. Koperski,
  "Object-Based Selective Materialization for
  Efficient Implementation of Spatial Data Cubes,"
  IEEE Transactions on Knowledge and Data
  Engineering, 12(6), 2000.