Notes on “Writing Successful Queries: Introduction to Predictive Search”
0:00 What is predictive searching?
--thanks to those who helped me name it
--Imagine what perfect source would look like use that to build search
--Karen Blakeman and Daniel Russel both wrote about it – as well as others
3:48 – How to do a Realtime search to find chats going on in Twitter (Note: Realtime is currently discontinued)
4:48 –Example of Predictive Search: I want to know who the two main actors are in Key Largo
--most obvious example for experienced internet users: IMDB
--but, basic predictive search can draw on other types of literacy, as well
--some people go to Images and see video cover or poster
--some go to Videos and look for clips/snippets
--some draw on experience as a reader—reviews and other articles start with phrases like “…the classic movie, Key Largo, starring…,” and so search for [key largo starring]
7:18 Could be called “Stop & Think”
--often hear that never occurs to people to stop and think before typing into a search bar
8:18 How Google Works
--Spiders crawl websites and records what is on the page
--Breaks into separate words and puts into index
--Every word goes into the index
--Compare searches for [who], [the who], [a who]—see get different answers
*Small words matter—helps us get at what you really want
10:58 So, if you search for [my cow has blisters on its tongue, what’s wrong with it?]
--Small words dilute the meaning you need
11:48 Student types in successful question: [what year did the american revolution start?]
--Predictably, gets answer sites
--Because once Google retrieves matching pages, ranks them by over 200 signals, including:
*words in title
*words in same order as you enter them
*words close together
--since answer sites put questions as the title of the page, tend to rank answer sites as good match
13:18 Big idea: You need to search for your answers, not your question
Google is written for everyone to use, not specifically for an academic audience, sometimes need to employ extra tools to find academic content.
14:13 How fast can a cheetah run?
One big challenge is that humans are so diverse and creative, so we express one idea in many different ways.
So, I think about the terms that others would use answering the question:
[cheetahs can reach speeds of up to mph]
--these are all words I can imagine being on a page with my answer
--plenty of results with this phrase
--can even add quotation marks to glue terms together: [“cheetahs can reach speeds of up to mph”]
*but missing a key element—the number of MPH
*so, can use this search: [“cheetahs can reach speeds of up to * mph”]
* the asterisk (*) creates a “fill in the blank”
16:18 But you need to think of the actual phrase that is likely to express the content you want.
So, when you want to know something about acceleration of a car, you need to draw on the common language/terms of art for automobiles
--common phrase heard in all car commercials: “0 to 60 in * seconds”:
*[2007 ford escape “0 to 60 in * seconds”]
*this does not draw on knowledge of the web, just having watched TV, or
seen auto ads in newspaper or magazine
18:08 More academic: looking for information on tooth decay in children
--[tooth decay children facts] or [tooth decay children information] a little fuzzy
*these extra search terms not as much in academic sources
*however, table and figures are common words in academic sources
*[tooth decay children tables OR figures]—finds better content (the OR in all
caps looks for either the word tables or looks for the word figures)
In fact, if you ask a student to think about where they would find numbers on a topic today, most will look for their search term and then go to Images and look for graphs.
Think about captions that go with the data they want.
Can search for something like [cardiovascular system bibliography] if what you want is a bibliography, can work well.
--Often, results will be .edu—may be too advanced for some K-12
--So: [cardiovascular system bibliography site:k12.*.us] will find on sites from school districts—across K-12 grades
25:18 What is that indentation above my lip?
--Need to know what it is called? Can use a casual-language search to find that answer.
--If you want formal information, use [philtrum] to find more technical information.
23:28 BUT, we see that students who learn to use terms of art start to pile on the most formal language they can think of and write very heavy sources
So, next level of skill—think about what kind of source will have the answer you need, and search with language that matches that type of source
--Son of a co-worker wanted to identify the song sung at Stanford basketball games with the lyrics “oh oh oh oh”
--Tried [oh oh oh oh song], turned out to be a perfect search because the type of source that had the answer was an answer site, and the language that people use on answer sites is very informal and highly descriptive
31:48 If you stop & think, may realize it is not an online search at all—may lead to a book, to a person, etc.
32:33 Also, not necessarily stay in open web: Heard that Napoleon had ongoing correspondence with wife Josephine,.
--Think that this would be most reliable in a book
--In fact, top search results for [napoleon correspondence josephine] is a book
--So, go to books.google.com (or click on Books in left-hand panel) and look for [napoleon correspondence josephine]
*Find the book, view the book (full text free online), or use left-hand panel to click on Find in a library link and locate it in a library near you
--Seemed natural to be in a book, and just had to think it through to get to the right place
34:55 Big idea: If I am not aware that some type of resource is possible, I would never think to use it—I might not even think to want a certain type of information if I am not aware that it exists.
35:38 For example: The current events of the day is Osama bin Laden’s death.
--If I don’t know that there are tools to help me do it, it might not even occur to me to desire to see how to read what people are saying about his death in Arabic, French, or Japanese.
--If I don’t know that there are tools to help me do it, I won’t know that I can ask other important points in history before 9/11 related to bin Laden.
39:03 How young can kids do this? I’ve had success from 3rd grade up.
41:15 Big Idea: Predictive Search is the flip side of AASL standard to know proper formats for communicating different kinds of information—instead, think about what it would look like, and search for it properly.
42:33 Kids working on a paper on childhood obesity—want statistics
[childhood obesity statistics] works, but if you want stats with commentary, more difficult
--forecasts for obesity can be describes as forecasts, predictions, estimates or many other search terms
--but imagine the sentence in the newspaper article, and realize has phrase “by 2015..2050” in it
--[childhood obesity “by 2015..2050”]
--similar: [diabetes “from * in 1940..1970 to * in 2005..2011”]
46:08 Big idea: Sometimes the keywords we want are messy, but there are associated terms in the nearby text that are fairly consistent. If we can visualize them, we can search for those instead.
47:00 Big idea: predictive search is used to write queries, but also to select the search result to click on—don’t look at one result at a time, look at page to see if your search is getting the type of sources you are anticipating, is there something wrong with your query, is there something wrong with the question you are asking?
Students did predictive search [different breeds of cats] –looking at results shows that they look good at first, but are actually quite messy.
--Results are informal, give varying answers that are difficult to access
--Results are not authoritative
--Results suggest that there is more to the question than the answers she is finding suggests