SLIDE 1IS 240 – Spring 2011 Prof. Ray Larson University of California, Berkeley School of Information Principles of Information Retrieval Lecture 21: Web Searching

SLIDE 2IS 240 – Spring 2011 Today Review –GIR and location-based Search Web Crawling and Search Issues –Web Crawling –Web Search Engines and Algorithms Credit for some of the slides in this lecture goes to Marti Hearst

SLIDE 3IS 240 – Spring 2011 Today Review –GIR and location-based Search Web Crawling and Search Issues –Web Crawling –Web Search Engines and Algorithms Credit for some of the slides in this lecture goes to Marti Hearst

SLIDE 4IS 240 – Spring 2011 Geographic information retrieval (GIR) is concerned with spatial approaches to the retrieval of geographically referenced, or georeferenced, information objects (GIOs) –about specific regions or features on or near the surface of the Earth. –Geospatial data are a special type of GIO that encodes a specific geographic feature or set of features along with associated attributes maps, air photos, satellite imagery, digital geographic data, photos, text documents, etc. Geographic Information Retrieval (GIR) Source: USGS

SLIDE 5IS 240 – Spring 2011 San Francisco Bay Area , Georeferencing and GIR Within a GIR system, e.g., a geographic digital library, information objects can be georeferenced by place names or by geographic coordinates (i.e. longitude & latitude)

SLIDE 6IS 240 – Spring 2011 Probability of relevance is based on Logistic regression from a sample set of documents to determine values of the coefficients. At retrieval the probability estimate is obtained by: For the m X attribute measures (on the following page) Logistic Regression

SLIDE 7IS 240 – Spring 2011 LR model X 1 = area of overlap(query region, candidate GIO) / area of query region X 2 = area of overlap(query region, candidate GIO) / area of candidate GIO Where: Range for all variables is 0 (not similar) to 1 (same)

SLIDE 8IS 240 – Spring 2011 Some of our Results Mean Average Query Precision: the average precision values after each new relevant document is observed in a ranked list. For metadata indexed by CA named place regions: For all metadata in the test collection: BUT: The inclusion of UDA indexed metadata reduces precision. This is because coarse approximations of onshore or coastal geographic regions will necessarily include much irrelevant offshore area, and vice versa

SLIDE 9IS 240 – Spring 2011 What can we do with GIR? Consider some earlier work on Mobile Media Metadata –Experimental development with Marc Davis during course I202 in 2003 and 2004 Several of the doctoral students currently in the program were participants in this course during that period –The following slides and video from Marc… Please recall that this was in so many of today’s web-based media technologies were still in early stages (or not developed, such as you- tube and Flickr)

SLIDE 10IS 240 – Spring 2011 What is the Problem? Today people cannot easily find, edit, share, and reuse media Computers don’t understand media content –Media is opaque and data rich –We lack structured representations Without content representation (metadata), manipulating digital media will remain like word-processing with bitmaps

SLIDE 11IS 240 – Spring 2011 Signal-to-Symbol Problems Semantic Gap –Gap between low- level signal analysis and high-level semantic descriptions –“Vertical off-white rectangular blob on blue background” does not equal “Campanile at UC Berkeley”

SLIDE 12IS 240 – Spring 2011 Signal-to-Symbol Problems Sensory Gap –Gap between how an object appears and what it is –Different images of same object can appear dissimilar –Images of different objects can appear similar

SLIDE 13IS 240 – Spring 2011 From Context to Content Context –When Date and time refined to qualitative semantic time periods (weekday, weekend, night, day, fall, spring, summer, winter) and recurring and special events –Where CellID (and GPS and geocoded Bluetooth access points) refined to semantic placenames –Who Cameraphone user Bluetooth co-presence Whom shared with Context

SLIDE 14IS 240 – Spring 2011 From Context to Content Context –When –Where –Who Content –Where is the location of the subject of the photo? –Who is depicted in the photo? –What objects are depicted in the photo? –What are the depicted people doing in the photo (i.e., actions, activities, events)?

SLIDE 15IS 240 – Spring 2011 MMM Revisited 2010 Technology has continued to develop and the social technologies have flourished Many of the Yahoo! Social media services including geotagging in Flickr have been influenced by MMM (Marc was the founding director of the Yahoo! Berkeley research lab) Many of the student MMM projects went on to become startups

SLIDE 16IS 240 – Spring 2011 But… The basic concept inspired by the Campanile example has not yet been fully realized In discussions on research collaborations with Dr. Satoshi Nakamura we have begun to define a related area combining –Photo-based GIR –Personal photo collections –Augmented reality mobile applications

SLIDE 17IS 240 – Spring 2011 Current Technology High speed network connections Built-in GPS –Time and geographic coords automatically embedded in photo metadata –Motion sensor embeds camera orientation and direction Development platform –Thousands of applications available Full computational capabilities Up to 32GB local storage

SLIDE 18IS 240 – Spring 2011 Augmented Reality Among the many applications available we have seen the rise of several Augmented Reality apps Basic concept –Use location information to Give directions Show current or historic data such as photos Leave messages at locations for others to find

SLIDE 19IS 240 – Spring 2011 Augmented reality - AcrossAir

SLIDE 20IS 240 – Spring 2011 Augmented reality - Sekai Camera

SLIDE 21IS 240 – Spring 2011 Augmented Reality

SLIDE 22IS 240 – Spring 2011 First Steps toward MMM3 Exploring the information available in the image metadata –Based on Dr. Nakamura’s GPS located photos taken during his trip to Berkeley –Also based on my iPhone pictures since getting a GPS enabled iPhone Using SQLite data generated by Dr. Nakamura’s lifelogviewer application Example and Demo of GIR

SLIDE 23IS 240 – Spring 2011 Yosemite Falls

SLIDE 24IS 240 – Spring 2011 Yosemite overview

SLIDE 25IS 240 – Spring 2011 Example source for Metadata

SLIDE 26IS 240 – Spring 2011 Other Collections and Metadata

SLIDE 27IS 240 – Spring 2011 Other Collections and Metadata From “Mapping the World’s Photos” David Crandall, Lars Backstrom, Daniel Huttenlocher and Jon Kleinberg Depar tment of Computer Science Cornell University

SLIDE 28IS 240 – Spring 2011 Next-Generation MMM (MMM3?) Continue to leverage the spatio-temporal context and social community of media capture in mobile devices –Gather all automatically available information at the point of capture (time, spatial location, phone user, other users, etc.) Add additional automatic metadata - camera orientation, direction of photo (can we triangulate objects based on multiple camera angles or analysis like MS photosynth) –Use metadata similarity and media analysis algorithms to find similar media that has been annotated before Exploit public resources such as Flickr, Panoramio, and Google Earth’s “points of interest”, etc. –Take advantage of this previously annotated media to make educated guesses about the content of the newly captured media As the number of tagged photos expands, guessing becomes easier –Interact in a simple and intuitive way with the phone user to confirm and augment system-supplied metadata for captured media Interact with other systems to support information, such as accrossair or Sekai camera –Personal library management and enhancement –Use social and image metadata to aid user in selection, presentation, etc. –Provide location memory for people

SLIDE 29IS 240 – Spring 2011 Today Review –GIR and location-based Search Web Crawling and Search Issues –Web Crawling –Web Search Engines and Algorithms Credit for some of the slides in this lecture goes to Marti Hearst

SLIDE 30IS 240 – Spring 2011 Standard Web Search Engine Architecture crawl the web create an inverted index Check for duplicates, store the documents Inverted index Search engine servers user query Show results To user DocIds

SLIDE 31IS 240 – Spring 2011 Standard Web Search Engine Architecture crawl the web create an inverted index Check for duplicates, store the documents Inverted index Search engine servers user query Show results To user DocIds

SLIDE 32IS 240 – Spring 2011 Web Crawling How do the web search engines get all of the items they index? Main idea: –Start with known sites –Record information for these sites –Follow the links from each site –Record information found at new sites –Repeat

SLIDE 33IS 240 – Spring 2011 Web Crawlers How do the web search engines get all of the items they index? More precisely: –Put a set of known sites on a queue –Repeat the following until the queue is empty: Take the first page off of the queue If this page has not yet been processed: –Record the information found on this page »Positions of words, links going out, etc –Add each link on the current page to the queue –Record that this page has been processed In what order should the links be followed?

SLIDE 34IS 240 – Spring 2011 Page Visit Order Animated examples of breadth-first vs depth-first search on trees: – Structure to be traversed

SLIDE 35IS 240 – Spring 2011 Page Visit Order Animated examples of breadth-first vs depth-first search on trees: – Breadth-first search (must be in presentation mode to see this animation)

SLIDE 36IS 240 – Spring 2011 Page Visit Order Animated examples of breadth-first vs depth-first search on trees: – Depth-first search (must be in presentation mode to see this animation)

SLIDE 37IS 240 – Spring 2011 Page Visit Order Animated examples of breadth-first vs depth-first search on trees:

SLIDE 38IS 240 – Spring 2011 Sites Are Complex Graphs, Not Just Trees Page 1 Page 3 Page 2 Page 1 Page 2 Page 1 Page 5 Page 6 Page 4 Page 1 Page 2 Page 1 Page 3 Site 6 Site 5 Site 3 Site 1 Site 2

SLIDE 39IS 240 – Spring 2011 Web Crawling Issues Keep out signs –A file called robots.txt tells the crawler which directories are off limits Freshness –Figure out which pages change often –Recrawl these often Duplicates, virtual hosts, etc –Convert page contents with a hash function –Compare new pages to the hash table Lots of problems –Server unavailable –Incorrect html –Missing links –Infinite loops Web crawling is difficult to do robustly!

SLIDE 40IS 240 – Spring 2011 Today Review –Geographic Information Retrieval –GIR Algorithms and evaluation based on a presentation to the 2004 European Conference on Digital Libraries, held in Bath, U.K. Web Crawling and Search Issues –Web Crawling –Web Search Engines and Algorithms Credit for some of the slides in this lecture goes to Marti Hearst

SLIDE 41IS 240 – Spring 2011 Searching the Web Web Directories versus Search Engines Some statistics about Web searching Challenges for Web Searching Search Engines –Crawling –Indexing –Querying Note: This is full of old data. The search companies are less forthcoming with exact numbers than they were a few years ago.

SLIDE 42IS 240 – Spring 2011 Directories vs. Search Engines Directories –Hand-selected sites –Search over the contents of the descriptions of the pages –Organized in advance into categories Search Engines –All pages in all sites –Search over the contents of the pages themselves –Organized after the query by relevance rankings or other scores

SLIDE 43IS 240 – Spring 2011 Search Engines vs. Internal Engines Not long ago HotBot, GoTo, Yahoo and Microsoft were all powered by Inktomi Today Google is the search engine behind many other search services

SLIDE 44IS 240 – Spring 2011 Statistics from Inktomi Statistics from Inktomi, August 2000, for one client, one week –Total # queries: –Number of repeated queries: –Number of queries with repeated words: –Average words/ query: 2.39 –Query type: All words: ; Any words: ; Some words: –Boolean: ( AND / OR / NOT) –Phrase searches: –URL searches: –URL searches w/http: – searches: –Wildcards: ( '?'s ) frac '?' at end of query: interrogatives when '?' at end: composed of: –who: what: when: why: how: where where-MIS can,etc.: do(es)/did: 0.0

SLIDE 45IS 240 – Spring 2011 What Do People Search for on the Web? Topics –Genealogy/Public Figure:12% –Computer related:12% –Business:12% –Entertainment: 8% –Medical: 8% –Politics & Government 7% –News 7% –Hobbies 6% –General info/surfing 6% –Science 6% –Travel 5% –Arts/education/shopping/images 14% (from Spink et al. 98 study)

SLIDE 46IS 240 – Spring 2011

SLIDE 47IS 240 – Spring 2011

SLIDE 48 Searches Per Day (2000)

SLIDE 49IS 240 – Spring 2011 Searches Per Day (2001)

SLIDE 50IS 240 – Spring 2011 Searches per day (current) Don’t have exact numbers for Google, but they have stated in their “press” section that they handle 200 Million searches per day They index over 8 Billion web pages – Now just says “Billions” - maybe they have stopped counting…

SLIDE 51IS 240 – Spring 2011 Challenges for Web Searching: Data Distributed data Volatile data/”Freshness”: 40% of the web changes every month Exponential growth Unstructured and redundant data: 30% of web pages are near duplicates Unedited data Multiple formats Commercial biases Hidden data

SLIDE 52IS 240 – Spring 2011 Challenges for Web Searching: Users Users unfamiliar with search engine interfaces (e.g., Does the query “apples oranges” mean the same thing on all of the search engines?) Users unfamiliar with the logical view of the data (e.g., Is a search for “Oranges” the same things as a search for “oranges”?) Many different kinds of users

SLIDE 53IS 240 – Spring 2011 Web Search Queries Web search queries are SHORT –~2.4 words on average (Aug 2000) –Has increased, was 1.7 (~1997) User Expectations –Many say “the first item shown should be what I want to see”! –This works if the user has the most popular/common notion in mind

SLIDE 54IS 240 – Spring 2011 Search Engines Crawling Indexing Querying

SLIDE 55IS 240 – Spring 2011 Web Search Engine Layers From description of the FAST search engine, by Knut Risvik

SLIDE 56IS 240 – Spring 2011 Standard Web Search Engine Architecture crawl the web create an inverted index Check for duplicates, store the documents Inverted index Search engine servers user query Show results To user DocIds

SLIDE 57IS 240 – Spring 2011 More detailed architecture, from Brin & Page 98. Only covers the preprocessing in detail, not the query serving.

SLIDE 58IS 240 – Spring 2011 Indexes for Web Search Engines Inverted indexes are still used, even though the web is so huge Most current web search systems partition the indexes across different machines –Each machine handles different parts of the data (Google uses thousands of PC-class processors and keeps most things in main memory) Other systems duplicate the data across many machines –Queries are distributed among the machines Most do a combination of these

SLIDE 59IS 240 – Spring 2011 Search Engine Querying In this example, the data for the pages is partitioned across machines. Additionally, each partition is allocated multiple machines to handle the queries. Each row can handle 120 queries per second Each column can handle 7M pages To handle more queries, add another row. From description of the FAST search engine, by Knut Risvik

SLIDE 60IS 240 – Spring 2011 Querying: Cascading Allocation of CPUs A variation on this that produces a cost- savings: –Put high-quality/common pages on many machines –Put lower quality/less common pages on fewer machines –Query goes to high quality machines first –If no hits found there, go to other machines

SLIDE 61IS 240 – Spring 2011 Google Google maintains (probably) the worlds largest Linux cluster (over 15,000 servers) These are partitioned between index servers and page servers –Index servers resolve the queries (massively parallel processing) –Page servers deliver the results of the queries Over 8 Billion web pages are indexed and served by Google

SLIDE 62IS 240 – Spring 2011 Search Engine Indexes Starting Points for Users include Manually compiled lists –Directories Page “popularity” –Frequently visited pages (in general) –Frequently visited pages as a result of a query Link “co-citation” –Which sites are linked to by other sites?

SLIDE 63IS 240 – Spring 2011 Starting Points: What is Really Being Used? Todays search engines combine these methods in various ways –Integration of Directories Today most web search engines integrate categories into the results listings Lycos, MSN, Google –Link analysis Google uses it; others are also using it Words on the links seems to be especially useful –Page popularity Many use DirectHit’s popularity rankings

SLIDE 64IS 240 – Spring 2011 Web Page Ranking Varies by search engine –Pretty messy in many cases –Details usually proprietary and fluctuating Combining subsets of: –Term frequencies –Term proximities –Term position (title, top of page, etc) –Term characteristics (boldface, capitalized, etc) –Link analysis information –Category information –Popularity information

SLIDE 65IS 240 – Spring 2011 Ranking: Hearst ‘96 Proximity search can help get high- precision results if >1 term –Combine Boolean and passage-level proximity –Proves significant improvements when retrieving top 5, 10, 20, 30 documents –Results reproduced by Mitra et al. 98 –Google uses something similar

SLIDE 66IS 240 – Spring 2011 Ranking: Link Analysis Assumptions: –If the pages pointing to this page are good, then this is also a good page –The words on the links pointing to this page are useful indicators of what this page is about –References: Page et al. 98, Kleinberg 98

SLIDE 67IS 240 – Spring 2011 Ranking: Link Analysis Why does this work? –The official Toyota site will be linked to by lots of other official (or high-quality) sites –The best Toyota fan-club site probably also has many links pointing to it –Less high-quality sites do not have as many high-quality sites linking to them

SLIDE 68IS 240 – Spring 2011 Ranking: PageRank Google uses the PageRank for ranking docs: We assume page p i has pages p j...p N which point to it (i.e., are citations). The parameter d is a damping factor which can be set between 0 and 1. d is usually set to L(p i ) is defined as the number of links going out of page p i. The PageRank (PR) of a page p i is given as follows: Note that the PageRanks form a probability distribution over web pages, so the sum of all web pages' PageRanks will be one

SLIDE 69IS 240 – Spring 2011 PageRank (from Wikipedia)

SLIDE 70IS 240 – Spring 2011 PageRank Similar to calculations used in scientific citation analysis (e.g., Garfield et al.) and social network analysis (e.g., Waserman et al.) Similar to other work on ranking (e.g., the hubs and authorities of Kleinberg et al.) How is Amazon similar to Google in terms of the basic insights and techniques of PageRank? How could PageRank be applied to other problems and domains?

SLIDE 71IS 240 – Spring 2011 Other Issues and Approaches Eric Brewer on Search Engine vs Database Technology: – acad/brewer