1. 1 Graphs & more on Web search 15-211 Fundamental Data Structures and Algorithms Stefan Niculescu & James Lyons March 21, 2002

2. Announcements

3. 3 Homework 5  Homework Assignment #5 will be out on Friday.  Must do some reading in order to complete it.  Must take a progress quiz.  Get started today and as usual, think b4 u hack!

4. 4 Reading  About graphs:  Chapter 14  About Web search:  http://www.cadenza.org/search_engine_terms /srchad.htm http://www.cadenza.org/search_engine_terms /srchad.htm  A HTML tutorial:  http://www.cwru.edu/help/introHTML/toc.html http://www.cwru.edu/help/introHTML/toc.html

5. Introduction to Graphs

6. 6 Graphs — an overview Vertices (aka nodes)

7. 7 Graphs — an overview PIT BOS JFK DTW LAX SFO Vertices (aka nodes) Edges

8. 8 Undirected Graphs — an overview PIT BOS JFK DTW LAX SFO Vertices (aka nodes) Edges

9. 9 Undirected Graphs — an overview PIT BOS JFK DTW LAX SFO Vertices (aka nodes) Edges 1987 2273 344 2145 2462 618 211 318 190 Weights

10. 10 Terminology  Graph G = (V,E)  Set V of vertices (nodes)  Set E of edges  Elements of E are pairs (v,w) where v,w  V.  An edge (v,v) is a self-loop. (Usually assume no self- loops.)  Weighted graph  Elements of E are (v,w,x) where x is a weight.

11. 11 Terminology, cont’d  Directed graph (digraph)  The edge pairs are ordered  Every edge has a specified direction  The Web is a directed graph  Undirected graph  The edge pairs are unordered  E is a symmetric relation  (v,w)  E implies (w,v)  E  In an undirected graph (v,w) and (w,v) are usually treated as though they are the same edge

12. 12 Directed Graph (digraph) PIT BOS JFK DTW LAX SFO Vertices (aka nodes) Edges

13. 13 Undirected Graph PIT BOS JFK DTW LAX SFO Vertices (aka nodes) Edges

14. 14 Terminology, cont’d  v and w adjacent (neighbors) if (v,w)E or (w,v)E  d(v) (degree of v) = # neighbors of v (for undirected graphs)  d + (v) (out-degree of v)= # of edges (v,w)E  d - (v) (in-degree of v)= # of edges (w,v)E

15. 15 Terminology, cont’d  Path a list of nodes (v[1], v[2],...,v[n]) s.t. (v[i],v[i+1])  E for all 0 < i < n  The length of the above path is n-1  Cycle a path that begins and ends with the same node  Cyclic graph – contains at least one cycle  Acyclic graph - no cycles

16. 16 Elements of a Graph PIT BOS JFK DTW LAX SFO

17. 17 Terminology, cont’d  Subgraph of a graph G a subset of V with the corresponding edges from E.  Connected graph a graph where for every pair of nodes there exists a sequence of edges starting at one node and ending at the other.  Connected component of a graph G a connected subgraph of G.

18. 18 Elements of a Graph, cont’d PIT BOS JFK DTW LAX SFO

19. 19 Terminology, cont’d  Unrooted (undirected) tree a acyclic connected undirected graph  Theorem: in any unrooted tree T=(V,E), |V|=|E|+1. Proof: by induction on |V|  Base case |V|=1 (|E|=0)  Show there exists a node of degree one  Remove that node and apply induction hypothesis

20. 20 Example of a unrooted tree PIT BOS JFK DTW LAX SFO

21. 21 Quiz Break

22. 22 So, is this a connected graph? Cyclic or Acyclic? Directed or Undirected? 6 4 5 3 1 2

23. 23 Directed graph (unconnected) Cyclic or Acyclic? 6 5 3 2 1 4

24. Representing Graphs

25. 25 Representing graphs  Adjacency matrix  2-dimensional array  For each edge (u,v), set A[u][v] to true; otherwise false 1 4 7 6 3 5 2 1234567 1xx 2xx 3x 4xxx 5xx 6 7 34 45 6 36 47 7  Adjacency lists  For each vertex, keep a list of adjacent vertices 1 2 3 4 5 6 7

26. 26 Choosing a representation  Size of V relative to size of E is a primary factor.  Dense: |E|/|V| is large  Sparse: |E|/|V| is small  Adjacency matrix is expensive in terms of space if the graph is sparse (O(|V| 2 > O(|E|+|V|)).  Adjacency list is expensive in terms of checking edges if the graph is dense.

27. 27 Size of a Graph  How many undirected graphs for a set of n given vertices?  Answer:  How many edges in a undirected graph with n vertices?  Minimum: 0  Maximum:

28. Graphs are Everywhere

29. 29 Graphs as models  The Internet  Communication pathways  DNS hierarchy  The WWW  The physical world  Road topology and maps  Airline routes and fares  Electrical circuits  Job and manufacturing scheduling

30. 30 Graphs as models  Physical objects are often modeled by meshes, which are a particular kind of graph structure. By Jonathan Shewchuk

31. 31 More graph models See also http://java.sun.com/applets/jdk/1.1/demo/WireFrame/index.html and http://www.mapquest.com NASA CFD labs By Paul Heckbert and David Garland

32. 32 Structure of the Internet Europe Japan Backbone 1 Backbone 2 Backbone 3 Backbone 4, 5, N Australia Regional A Regional B NAP SOURCE: CISCO SYSTEMS MAPS UUNET MAP

33. 33 Relationship graphs  Graphs are also used to model relationships among entities.  Scheduling and resource constraints.  Inheritance hierarchies. 15-11315-151 15-211 15-21215-213 15-312 15-41115-412 15-251 15-45115-462

34. 34 Where are we right now?

35. The Web Graph

36. 36 Web Graph  Documents written in HTML  HTML (HyperText Markup Language)  TAGS: ,, ,  (anchor, link)

37. 37 A simple HTML example A Simple HTML Example Carnegie Mellon University

38. 38 Web Graph  A directed graph where :  V = (all web pages)  E = (all HTML-defined links from one web page to another)

39. 39 Web Graph  Web Pages are nodes (vertices)  HTML references are links (edges)

40. 40 Is the Web Graph connected?  Sparse, unconnected graph  AUTHORITIES web pages containing a “reasonable” amount of relevant information about a specific topic  HUBS web pages that point (link) to many pages containing relevant information about a given topic

41. 41 Finding Hubs & Authorities  Nice iterative algorithm by Jon Kleinberg www.cs.cornell.edu/home/kleinber/auth.ps  HUB: Avrim’s Machine Learning page www.cs.cmu.edu/~avrim/ML  AUTHORITY: www.java.sun.comwww.java.sun.com  Extra credit opportunity for homework 5

42. Graphs : Application Search Engines

43. 43 Search Engines

44. 44 What are they?  Tools for finding information on the Web  Problem: “hidden” databases, e.g. New York Times (ie, databases of keywords hosted by the web site itself. These cannot be accessed by Yahoo, Google etc.)  Search engine  A machine-constructed index (usually by keyword)  So many search engines, we need search engines to find them. Searchenginecollosus.comSearchenginecollosus.com

45. 45 Did you know?  Vivisimo was developed here at CMU  Developed by Prof. Raul Valdes-Perez  Developed in 2000

46. 46 SE Architecture  Spider  Crawls the web to find pages. Follows hyperlinks. Never stops  Indexer  Produces data structures for fast searching of all words in the pages (ie, it updates the lexicon)  Retriever  Query interface  Database lookup to find hits  1 billion documents  1 TB RAM, many terabytes of disk  Ranking

47. 47 A look at  10,000 servers (WOW!)  Web site traffic grows over 20% per month  Spiders over 2 Billion URLs  Supports 28 language searches  Over 100 million searches per day  “Even CMU uses it!”

48. 48 Google’s server farm

49. 49 Web Crawlers  Start with an initial page P 0. Find URLs on P 0 and add them to a queue  When done with P 0, pass it to an indexing program, get a page P 1 from the queue and repeat  Can be specialized (e.g. only look for email addresses)  Issues  Which page to look at next? (Special subjects, recency)  How deep within a site do you go (depth search)?  How frequently to visit pages?

50. 50 So, why Spider the Web?  Refresh Collection by deleting dead links  OK if index is slightly smaller  Done every 1-2 weeks in best engines  Finding new sites  Respider the entire web  Done every 2-4 weeks in best engines

51. 51 Cost of Spidering  Spider can (and does) run in parallel on hundreds of severs  Very high network connectivity (e.g. T3 line)  Servers can migrate from spidering to query processing depending on time-of-day load  Running a full web spider takes days even with hundreds of dedicated servers

52. 52 Indexing  Arrangement of data (data structure) to permit fast searching  Which list is easier to search? sow fox pig eel yak hen ant cat dog hog ant cat dog eel fox hen hog pig sow yak  Sorting helps. Why?  Permits binary search. About log 2 n probes into list  log 2 (1 billion) ~ 30  Permits interpolation search. About log 2 (log 2 n) probes  log 2 log 2 (1 billion) ~ 5

53. 53 Inverted Files A file is a list of words by position - First entry is the word in position 1 (first word) - Entry 4562 is the word in position 4562 (4562 nd word) - Last entry is the last word An inverted file is a list of positions by word! POS 1 10 20 30 36 FILE a (1, 4, 40) entry (11, 20, 31) file (2, 38) list (5, 41) position (9, 16, 26) positions (44) word (14, 19, 24, 29, 35, 45) words (7) 4562 (21, 27) INVERTED FILE

54. 54 Inverted Files for Multiple Documents DOCID OCCUR POS 1 POS 2...... “jezebel” occurs 6 times in document 34, 3 times in document 44, 4 times in document 56... LEXICON WORD INDEX

55. 55 Ranking (Scoring) Hits  Hits must be presented in some order  What order?  Relevance, recency, popularity, reliability, alphabetic?  Some ranking methods  Presence of keywords in title of document  Closeness of keywords to start of document  Frequency of keyword in document  Link popularity (how many pages point to this one)

56. 56 Spamdexing & Link Popularity  Spamdexing means influencing retrieval ranking by altering a web page. (Puts “spam” in the index)  www.linkpopularity.com www.linkpopularity.com  Link popularity is used for ranking  Many measures  Number of links in (In-links)  Weighted number of links in (by weight of referring page)

57. 57 Search Engine Sizes AVAltavista EXExciteFAST GGGoogle INKInktomi NLNorthern Light SOURCE: SEARCHENGINEWATCH.COMSEARCHENGINEWATCH.COM

58. 58 Historical Notes  WebCrawler: first documented spider  Lycos: first large-scale spider  Top-honors for most web pages spidered: First Lycos, then AltaVista, then Google...

59. 59 Overview  Engines are a critical Web resource  Very sophisticated, high technology, but secret  Most spidering re-traverses stable web graph  They don’t spider the Web completely  Spamdexing is a problem  New paradigms needed as Web grows  What about images, music, video?  Google’s image search engine  Napster