INF 385D - Fall 2004
Wired Research Paper: Advanced Search
Valerie Gomez de la Torre

Introduction

The goal of a user when using a search engine is to find information about a specific topic in a fast and efficient manner. A search engine attempts to satisfy this goal by providing the user with a tool that tries to retrieve and return the most relevant results as quickly and simply as possible. While this seems like a win-win situation, users are still not using these search engines to their full capacity and search engines are not educating their users on how to create better queries. Search engines provide a basic search and an advanced search. They attempt with the basic search to return results as relevant as possible through a variety of algorithms and indexing practices. When these results do not satisfy a user, the search engine provides options on how to restrict a query so as to provide better results. The problem arises in that users either do not know how to access or use these more powerful search options.

This paper attempts to explain what users need to know in order to successfully use search engine´s advanced search options. It will also provide insight as to why these advanced options are not being used frequently by users and some suggestions on how to make them more accessible.

1. What is Advanced Search?

"Advanced search" is an added feature offered by most search engines to provide users with the option to refine the scope of and/or limit their search. If a user understands the principles of the advanced search options and how to use them, they can increase the relevancy of their search results. Below are four ways a user can use a search engine´s advanced options to restrict their query:

• Boolean Logic
• Implied Boolean Logic
• Proximity Operators
• Search Template/Advanced Interfaces

1.1 Boolean Logic

Boolean logic is a system written by British mathematician George Boole (1815-1864), which was designed to produce better search results by referring to the logical relationships among search terms. Boole called it the "calculus of thought".""This system is composed of the operators: OR, AND, and NOT, which are used to "link words and phrases for more precise queries" (Cohen, 2002).

Linda Cohen gives an excellent tutorial using easy to understand Venn diagrams to explain how Boolean logic works. Below are examples she offers:

• OR

The operator "OR" is often used to combine synonyms or like concepts.

The example above displays results from a query such as: I would like information about college. In this search a users will receive results with AT LEAST ONE of the search terms on the page. "The shaded circle with the word college represents all the records that contain the word ´college´. The shaded circle with the word university represents all the records that contain the word ´university´. The shaded overlap area represents all the records that contain both ´college´ and ´university´" (Cohen, 2004).

• AND

The "AND" operator is used to retrieve documents with all queried words in the results.

The diagram above represents a query such as: I want information relating to both poverty and crime. This query would return pages in which BOTH of the search terms are present. "This is illustrated by the shaded area overlapping the two circles representing all the records that contain both the word ´poverty´ and the word ´crime´. Notice how we do not retrieve any records with only ´poverty´ or only ´crime´" (Cohen, 2004).

• NOT

The "NOT" operator informs the search engine to retrieve pages that contain the first word queried where the second term is not present.

The diagram above shows the results of a query such as: I want pages with information about cats, but that have nothing to do with dogs. In this search, a user would receive results in which ONLY ONE of the terms is present. "This is illustrated by the shaded area with the word cats representing all the records containing the word ´cats´. No records are retrieved in which the word ´dogs´ appears, even if the word ´cats´ appears there too"(Cohen, 2004).

1.2 Implied Boolean Logic

Implied Boolean logic refers to a search in which symbols are used to represent Boolean logical operators. By using special characters and operators, such as +, -, and "" a user can fine-tune their search query and increase the accuracy of its results.

According to Danny Sullivan of SearchEngineWatch.com, "search engine math is much easier, conceptually, for users to understand" (Sullivan, 2001. In this type of search even the lack of a symbol is significant. For example, a search engine may recognize a space between keywords as either OR logic or AND logic. While many search engines would default to using the OR logic, there has been a movement towards using more the AND logic s a default. "Implied Boolean logic has become so common in Web searching that it may be considered a de facto standard" (Cohen, 2004).

Examples of implied Boolean logic:

• +

If a common word is essential to getting the best results, placing a "+" sign in front of it will force the search engine to include the word in the search results. For example, if a user wanted to search for pages containing information about BOTH Angelina Jolie AND Billy Bob Thornton, they would perform the following query:
+jolie+thornton

• -

This operator offers the opposite search refinement as the "+". For example, a user may want to search using a word that may have more than one meaning; "Paris" can refer to the city or to Paris Hilton, the person. With the "-" a user can instruct the search engine to avoid the word following the operator as in the following query:
-: pairs-hilton

• ""

Using quotation marks will retrieve pages where the terms you query appear in exact order a user specified. For example a user may want to find sailing reservations in Galveston. You could use the "+" operator such as +galveston+sailing+reservations. But this may return documents with Galveston in the title, sailing within the copy and reservations for another sport. A user can improve on their result by simply putting quotation marks around the phrase, like this:
"Galveston sailing reservations"

1.3 Proximity Operators

Proximity operators are used to instruct the search engine to search for documents that contain the queried words that are within a short distance of each other in a document. "For example, using a search engine that supports proximity operators, querying the phrase "cable NEAR modem" will instruct the search engine to look in documents for instances of the words "cable" and "modem" that are near each other" (Webopedia).

Some examples of proximity operators are:

• NEAR: searches for terms that are located in a document within a specific distance of each other in any order
• ADJ: "adjacent to", similar to the phrase searching except that the terms may appear in any order
• SAME: terms found in the same field
• FBY: "followed by", not yet used with search engines. Mainly used in library and specialized databases.

1.4 Search Template

For users who do not want to remember which operators to use they can fill out the top portion of the Advanced Search form.

Figure 1

The search form (see Figure 1) offers input areas with the following suggestion on how a user may want to restrict their query:
1. "with all of the words" This is the same as if the user were to submit a query using the "AND" operator
2. "with the exact phrase" This is the same as if the user were to submit a query using the "" operators
3. "with at least one of the words" This is the same as if the user were to submit a query using the "OR" operator
4. "without the words" This is the same as if the user were to submit a query using the "NOT" operator

2. Do users use advanced search?

A study done by Spink, Jansen, Wolfram, and Saracevic (2001) analyzing more than 200,000 queries submitted to the Excite search engine showed that while the numbers have improved over the years (see figure 2), users are still not taking advantage of the advanced search options offered by search engines.

Figure 2

A few reasons have been posed for why users are not using and benefiting from more advanced searching options:

• The label "advanced search" can be intimidating
  "Advanced" might turn off users who don't consider themselves advanced. Or it might conjure an image of a complicated interface that not useful enough to bother learning" (Rosenfeld).
• The functionality is not located in the proper place
  "When do we really need "advanced" functionality? When we want to revise a search that didn't perform as well as we'd hoped. When does that happen? After performing that initial search, naturally. Instead of the false dichotomy of "Simple/Normal/Basic Search" OR "Advanced Search", it seems to make sense to lead with a simple search box and then provide a "Revise Search" capability for users who aren't happy with the results of their simple search" (Rosenfeld).

3. How can we make advanced search better?

Applying information architecture principles such as proper labeling and grouping actions in the proper context may provide the benefits of the advanced search to users in a less intrusive or intimidating way.

• Change the label and the language

  "´Advanced search´ could become ´revise search´ or ´refine search´. Revising the search means repeating back the search query and enabling users to broaden or narrow its scope (e.g., by adding new terms from a thesaurus) before re-executing it"(Rosenfeld, 2002).

  Evidence has been found that users are put off by the terms ´advanced search´ or ´expert search´ and seldom read ´hints and tips´, believing these functions to be for experienced users (Hockley, 1997).

• Provide the "Revise" function on the results page

  Users would benefit more by having the option to revise a search that didn't perform as well as they had hoped after performing their initial search. "It seems to make sense to lead with a simple search box and then provide a "Revise Search" (See Figure 3) capability for users who aren't happy with the results of their simple search" (Rosenfeld, 2002).

Figure 3.

4. Conclusion

The more a user knows about how a search engine and its features work, the better it can serve their needs. This will also benefit the commercial search engines as "advertisers are ready to pay royally to reach this type of targeted audience" (Business Week Online).

While it is fair to assume that users will still cry, "don´t make me think" as stated by Steve Krug, search engines developers are beginning to "upport users, who, out of frustration, have become more search-literate and are willing to spend more time learning a complex search interface and constructing a query" (Rosenfeld and Moreville, 2002. p 165.). One thing to keep in mind - there will always be novice, inexperienced users using the internet to search for answers.

5. References

Cohen, Linda. University Libraries. May 2004. University of Albany. Retrieved:28 Nov. 2004. http://library.albany.edu/internet/boolean.html

Elgin, Ben, Jay Greene, and Steve Hamm. "The future of search". Business Week Online. 3 May 2004. Retrieved: 4 Nov. 2004. http://www.businessweek.com/magazine/content/04_18/b3881006_mz001.htm.

Hockley, Andrew, Annabel Pollock. "What's wrong with internet searching". Mar. 1997. Human Factors Unit, BT Laboratories. Retrieved: 2 Nov. 2004. http://www.dlib.org/dlib/march97/bt/03pollock.html.

Krug, Steve. (2000). Don't make me think. 1st ed. Indianapolis, IN: New Riders

LouisRosenfeld.com. "IA Heuristics for Search Systems".16 Dec. 2002. Retrieved: 1 Nov. 2004 http://www.louisrosenfeld.com/home/bloug_archive/000290.html

Rosenfeld, L., & Morville, P. (2002). Information architecture for the World Wide Web. 2nd ed. Sebastopol, CA: O'Reilly

Spink, A., Jansen, B. J., Wolfram, D., and Saracevic, T. 2002. From E-sex to E-commerce: Web Search Changes. IEEE Computer. 35(3), 107 - 111. View in PDF

Sullivan, Danny. Search Engine Watch. "Search Engine Math". 26 Oct. 2001. Retrieved:28 Nov. 2004 http://searchenginewatch.com/facts/article.php/2156021