For any online database there are two parts: the data set and the interface for accessing the data. Different commercial vendors make data sets available and create an interface or search engine for accessing the data. Some vendors supply many different data sets via the same interface while some interfaces are unique to a given data set. The Advanced Search Space of these different interfaces share certain functionalities which are described below.
Exercise no. 1: for all examples shown below, there are exercises which require you to access a specific database in order to practice a technique or strategy. In order to do so you must go the the Research Databases List maintained by Kelvin Smith Library. Your first exercise is to go to the webpage for the Kelvin Smith Library (library.case.edu) and find a link which takes you to the Research Databases list. Note briefly how it is set up and and how you use it to access a specific database.
Most database interfaces include a simple search function which consists of a single box for entering search terms which will then be searched as keywords in most of the fields in the database records. The advanced search function has muliple boxes which allow you enter keywords for searching, but allow you to specify which database index is searched. Many interfaces for academic databases open in the advanced searching mode, but some open to the simple search function.
Exercise no. 2: access the following the following databases: RILM Abstracts of Musical Literature, UMI ProQuest Digital Dissertations, and WorldCat, and notices the similarities and differences in the Advanced Search Space for the three different interfaces.
The Advanced Search Space is at its most basic a keyword searching space. The three fundamental techniques in keyword searching are: logical connectors (AND, OR, NOT), term versus phrase searching, and special techniques such as truncation of a term or the insertion of a wildcard character in a term.
When more than one keyword is entered in a search box, you must establish a “logical” relationship between the terms, which is done by connecting the terms with a logical connector such as AND, OR, NOT. Terms connected by AND will retrieve citations containing both terms (Mozart AND ornamentation), terms connected by OR will retrieve citations containing either term (orchestration OR instrumentation), and terms connected by NOT will retrieve citations containing the first term but exclude citations if the second term is also present (Haydn NOT review). Within the Advanced Search Space you may use logical connectors within a search box or between search boxes.
Long strings of keywords connected by different logical operators will often not work in expected ways, since more than one type of logical operator requires that you group terms within parentheses. If you tried to construct a search to find materials on the influence of Haydn on Mozart and Beethoven, most database interfaces will interpret:
Haydn AND influence AND Mozart OR Beethoven
as: (Haydn AND influence AND Mozart) OR Beethoven
and retrieve many citations which contain only “Beethoven”.
Depending on the database interface you will have to add parentheses to make the intended logical relationship clear or use multiple search boxes to do the same:
One of the reasons Advanced Search Spaces have multiple boxes is to address this problem.
To make the results of your searching more predictable, try to ascertain how a given interface handles logical operators and strings of keywords. Generally databases treat strings of keywords as if connected by AND. If you want terms connected by OR or NOT, be explicit and if you want a several terms searched only as a phrase, use quotes.
Exercise no. 3: In the database WorldCat search: music education. Then search: music and education. Note the total number of records retrieved. If they are the same number (or very close) then it signifies that WorldCat treats strings of keyword as they they were connected by AND. You have moved a step closer to a transparent understanding of that database.
You notice in the example above about clarifying logical operators that quotes were used to inclose "influence on". Keyword indexes can search for individual terms or phrases. The use of quotes generally guarantees a phrase search.
Exercise no. 4: In the database ERIC search: music education. Then search: "music education". Note the number of results obtained. If there are fewer retrieved results with the use of quotes, which is logical since the terms music and education may occur in a record when they are not in a phrase. You have moved a step closer to a transparent understanding of that database.
The example below shows in a database with a huge number of records, how significant term versus phrase searching can be.
Two specialized symbols which are functional in most databases are the wildcard and truncation symbols. The wildcard (which can vary between interfaces) allows you to place an unknown character within a word: a search using wom?n will retrieve records with woman or women. The truncation symbol (which can vary between interfaces) allows you to place an unknown character(s) at the end of a word: a search on instrument* will retrieve records with instrument, instruments or instrumentation. These symbols are one way to expand a simple keyword search, but you should check the help screens within a given interface to confirm which characters are used to represent these concepts.
Exercise no. 5: In the database UMI ProQuest Digital Dissertations search in the Document Title index: music*. Then search again on: music. If the search with the asterisk retrieves more records, than it probably found records with the term: musical.
Note that the concept of truncation of a term comes up without the truncation symbol. Most notably, online library catalog support truncation without any truncation symbol. If you do an author search on: Bach, J. You will be placed in the part of the alphabet of authors where Bach, J would occur and you can select the correct author heading and proceed.
Exercise no. 6: In the CWRU online library catalog (catalog.case.edu) do a Subject search on any composer of interest to you. Do so in the form: last name, first initial. Besides the entry for the composer's name as a subject (general works on the composer), note all the subject headings with sub-headings. Whenever an online database functions in this way it is said to support truncation. When you approach a new database for the first time, you should be able to answer the question, "does this database support truncation" with a single search.
Two specialized logical operators which are functional in most databases are proximity indicators or adjacency terms. These are generally labeled N (near) and W (within). They join two words in an AND relationship but specify a range of words which may be present between the two words. Searching on “Dresden N3 Opera” as a title will retrieve records with titles which include "Dresden" and "opera" near each other by no more than three words: "Dresden Opera", "opera in Dresden", and "Dresden archives include opera scores". Such a search would not retrieve a record with an abstract which included: "... fire in Dresden but the archives housed in the opera house ...". The W term works in a similar fashion, but restricts the adjacency to the same order in which you entered the terms. “Dresden W3 Opera” would retrieve a record with a title which included: "Dresden archives include opera scores", but not one with the title: "Opera in Dresden". Proximity indicators are one way to narrow a search on two keywords, but you should check the help screens within a given interface to confirm which characters are used to represent these concepts.
Most searching should not require these specialized logical operators, however they can be useful if your searchs are retrieving large numbers of database records and other keyword searching techniques are not effective in reducing the number of records retrieved while still retrieving relevant records.
Exercise no. 7: In the database PsycINFO search: children AND creativity. Then search: children N5 creativity. Note the number of results. The N5 search finds the two terms within 5 words of each other in any order. If the proximity search retrieves significantly less, than you have probably retrieved fewer records or greater relevance. You should have found records which include: creativity and children (as a phrase), creativity in children, children and creativity (as a phrase), but not records where the term children appears in the title and creativity appears in the abstract.
One of the most important features of the Advanced Search Space is the ability to select a specific keyword index for searching. Most of these indexes are labeled in such a way as to make it obvious what fields in the database record they search. It is obvious what keyword index labeled Author covers. But what if it is labeled Source Title, or Named Person, or Document Type, or Methodology? What if there are two keyword indexes, one called Author and the other Author Phrase? The best way to understand a specific keyword index is to look at a database record.
Exercise no. 8: in the database RILM, look at all the indexes available in the dropdown menu next to a search box. Note there is a index called: Source. What is this for? There is also one called Journal Title, so what is Source? Do a general keyword search on: music. Look at several full displays for database records retrieved and you will observe Source is whatever source the database record describes: a journal, a book, a dissertation, a conference proceeding. We can rightly assume a keyword search on Source will look through all sources, regardless of type. This implies that the index called Journal Title is more limited and will only search sources which are journals.
One of the major steps toward making a database appear more transparent is understanding the relationship between the name of a keyword index and the parts of a database record which it indexes. This understanding includes more than the connection between the index and the database record as shown in the next exercise.
Exercise no. 9: in the database UMI ProQuest Dissertations and Theses execute a general keyword search on music. Pick any record retrieved and choose Citation/Abstrct. Evaluate the abstract and the subjects. How detailed is the abstracted? How deep is the subject indexing, that is how specific is it and how many aspects of the original items does it cover? You should notice that this database has one of the longest and most detailed abstracts of any database, while it has some of the most superficial subject indexing of any database. This means that when you use the Abstract keyword index you may (should) should very specific terms and many of them in your search. The Subject Heading index will be of almost no use for most topics.
In the example below a RILM record has been analyzed in terms of how the database fields related to various indexes:
The default or general keyword index used in a database is the one selected when you first open the Advance Database Function. It may be labeled Select a Field (optional) as in RILM, or Keyword as in WorldCat, or All Fields as in UMI Proquest Digital Dissertations and Theses. You should always ascertain quickly which fields it is actually covering and the best way to do this is to look in the database's help screens, although locating this information can be a challenge.
Exercise No. 10: in the database International Bibliography of Theatre & Dance with Full Text locate the help button in the top bar upper right. Once in help scroll all the way to the bottom for database help specific to this database (and not the others that this vendor provides). You will find your answer to what the general (or unqualified) searching covers. Knowing this will inform your decisions about when to use the specific keyword indexes and when to use the general keyword index.
An important facet of most Advanced Search Spaces is the ability to add search limits which can be applied before your searches are executed. The most straightforward way to make searching on strings of keywords more precise is add some of these limits to a search. A useful set of limits for many patrons would be limiting by language and document type as in the example below.
The specific data set of a given online database will determine what types of limiters are available. In the example below, the nature and structure of the data set allows limits specific to that data set:
Note that in the Advanced Search Space, one of the reasons for multiple search boxes is so they can function as limiters. Note the example below in which searching has been limited to a specific journal title:
All of the examples above show pre-search limiting, but limiting can usually be done before or after a search. A current example of post-search limiting is the sidebar which displays in some interfaces after you have executed your search:
While the sidebar format types in this example are straightforward and their use should produce predictable results, the subject limiters represent only some of the subject headings attached to these records. Even clicking on the "Show More" does not give all subject headings associated with the retrieved records. As such these subject limiters can be very helpful in limiting a search quickly, but they also give the user a false impression of the subject headings which may be available for refining a search. The sidebar limiters based on physical format (journal articles) or dates will give more predictable results.
Exercise no. 11: in the database Business Source Complete search on: copyright and music. Note number of results. Add the post-limit search of Full Text. Note the number of results. Limit again by Peer Review. Note the number of results. This is an excellent way to begin to understand the content of a database.
One of the most effective forms of post-search limiting is adding terms to your original search and reexecuting the search.
Exercise no. 12: in the database MEDLINE with Full Text in one search box put: musicians and injuries. Execute the search and note number of records retrieved. In another search box add: repetitive. Reexecute the search and note the number of results. You have effectively massaged your results of article on musician's injuries to limit it to those of a repetitive nature. This is an excellent way to start to evaluate the content of a database.
The next level of expertise in using various interfaces for online databases is to understand and use the interface’s browse indexes. These sometimes function as feeder workspaces where you select terms or phrases and feed them back to the Advanced Work Space. Other times they act like parallel work spaces where you select terms or phrases and do your searching there in the browse index.
There are two types of browse indexes: term and phrase. The term indexes are based on individual keywords extracted from a field, such as keywords extracted from subject headings. The phrase indexes are often phrases from fields which are based on standardized entries such as author names or journal titles. The value of indexes is twofold. First, they can save you time and aggravation trying to figure out if the terms, phrases or names you use when constructing your searches are actually in the database. Second, they segregate different types of terms or phrases so you are searching on smaller, more uniform groups of terms or phrases. Browse ndex searching is more precise than keyword index searching.
Browse indexes allow you to see the list of terms or phrases contained in a given index. This functionality is usually labeled “browse” or “browse indexes” or in some cases just “indexes”. This functionality is one of the main reasons to have indexes -- the ability to see what is in the indexes makes the database much more transparent to the user. Different interfaces use different procedures for accessing the browse indexes: Note that a browse index may contain single terms or phrases. In the example below, we are checking the correct form of an organization (to later search for their publications). In WorldCat the browse indexes are accessed through the button to the right of the box listing the index.
Exercise No. 13: in the database WorldCat enter the search: Hefling, Stephen in the Author Phrase index and then click on the browse button. Note what you get and why this is a useful technique. Go back to the Advanced Search Space and reenter the search but use the Author index instead. Note the difference -- one is a (single) term index and the other is a phrase index. It is important to understand when a browse index is a term index and when it is a phrase index. It is easy to find out -- you just did.
The browse indexes in RILM are simply called "Indexes". Note the example below. You must choose the logical operator (AND or OR) when you have chosen several terms from the list and wish to add them to the search box. Note how the interface returns terms to the search box but does so in an advanced command language. You can learn to use this language directly, but it is not necessary in order to develop excellent searching skills. It is sufficient that you understand what it is, so that the searches you create by these kinds of techniques make sense to you.
Exercise no. 14: in the RILM, find the browse indexes (top bar, under more). As your browse index select Major Topic. Since you are not sure what this is, search on: abc. This will get you into the start of the index. Browse forward until you get to: Performance practice and notation -- performance practice, ca. 1800-1900. Click the box in front of this heading and then click ADD. Your choice will appear in the search box at the top. To what is in the search box, add: AND violin. Execute the search. If successful you have search for violin performance practice during the 19th century in RILM in the most effective manner possible.
The thesaurus is a specialized type of subject heading list in which the subject headings are listed in a manner which clarifies the relationship between different subject headings. These clarifications include notes which relate terms in various ways, indicating which are the preferred headings (lute USE FOR: theorbo, chittarone) (theorbo USE: lute), or describe the hierarchy of terms (lute BROADER TERM: stringed instruments) (string instruments NARROWER TERMS: lute, guitar, etc.). For some databases, the thesaurus is constructed and sanctioned by a professional organization in that subject field (as is the case with some of the psychology databases). A thesaurus is a controlled vocabulary for a given subject area or discipline and its value lies in the precision of its terminology and the precision with which indexers (real live people) apply the terms when indexing citations. Use of the thesaurus is particularly important if you are searching in a field which is not your main academic area and for which you may be less familiar with the standard vocabulary for that field. The search example below might be a first step in researching the relationship between right/left brain dominance and musical ability:
The thesaurus usually allows you different choices for searching a term. These choices will retrieve a greater or lesser number of citations as indicated by their labels: focus, search, expand. You need to look at a specific interface’s or database’s help screen to understand precisely what is being focused or expanded in a search. Possibilities include whether a subject heading appears in a record as a primary or secondary subject heading, or whether narrower terms for a given subject heading are or are not searched. The thesauri in the humanities databases are often not as elaborate as those in the social sciences, but they still can help guide you through the maze of authorized subject headings:
In the example above, checking the explode mechanism indicates you wish to search the main term (MOTION picture music) plus all narrower terms which the thesaurus lists under that term (such as ANIMATED film music). This type of funtionality is typical of most thesauri and is based on an important principle of indexing: items represented by citations are indexed at the most specific level applicable to the item. A book on animated film music is indexed under that heading, "Animated film music" but not under "Motion picture music". The explode function allows you to broaden your search by including more related thesaurus terms.
Thesauri (and other types of lists within databases which attempt to list subject phrases in a logical structure) are very powerful tools but difficult to understand when first encountered. For any database you expect to use frequently, you should experiment with the thesaurus or subject headings list until you feel you have a good grasp of how they work.
Exercise no. 15: in the database Art Full Text find the thesaurus. Search in the thesaurus for: Term contains music. Once you get results click on the heading Art & music. Examine the thesaurus record for Art & music. Click the boxes for Art & music as well as Music in art. Click on ADD (retaining OR as the logical connector). Once the search appears in the search box, execute the search. Examine your results to see if they were what you expected.
Most interfaces store searches you have executed as part of a search history of your session in the interface. These saved searches can be treated as adjunct workspaces which allow you to concentrate on different facets of your search in turn, maximizing the results for each individual facet, and then allowing you to bring the fully developed facets together in a final search. In the example below you can see how the use of search histories expands the usefulness of a typical type of searching behavior -- finding a relevant record via keyword searching, and then clicking on a standard subject heading. The only way to easily combine searches executed from clicking on different subject headings is to access your search history and combine sets.
Search histories are particularly important when using a thesaurus. Often you will want to find the different facets of your topic in a thesaurus separately, execute separate searches on each and then bring those searches together by combining sets in your search history.
Scholars cite the works of other scholars in footnotes, endnotes, and bibliographies. These references point to scholarship that supports or is related to their own. Obviously you know who has been cited by the author of the article or book which you have just read, but discovering who is citing that author or that specific work is another matter. Some databases link citations in such a way that you can discover, at least within that database, who is citing a given author or a specific work by a given author. This can exist as a function or specialized workspace within an interface/database (e.g. EBSCO interface and Humanities International Complete database) or it can be the basis for an interface/database (e.g. ISI Web of Knowledge interface and the Arts and Humanities Citation Index database). In the example below note how you have to identify the author who is cited and the work or works by that author which were cited before you can find any citing articles.
The search above shows how to find works which cite a work you have read. Some databases will also lead you to “related records”. Generally "related records" are items which share a reference with the work you have read -- in other words, which works cite the same sources as the author you are interested in? Related records results will generally give you many more citations, be less relevant, but take you further into the web of scholarship which surrounds the work you have read.
The example below shows a simple search in the Arts & Humanities Citation Index. This database was constructed as a citation database and is very powerful, but rather difficult to use. The search attempts to find works which cite a work by Case Music Professor Georgia Cowart.
A logical outcome of vendors providing a standardized interface which accesses different data sets is the ability to search more than one database simultaneously. The mechanism is simple: you access one database and then choose the others you wish to search simultaneously. In the example below we have opened RILM which is an EBSCO database. By clicking on "Choose Databases" you can select other EBSCO databases and search on all simultaneously.
The drawback to this procedure is the loss of indexes not shared by multiple databases. In the example below note the difference in indexes available in RILM and then the indexes available when RILM and MLA are searched together:
EBSCO offsets the loss of indexes, which in some databases is a very significant loss, by listing separately the pre-search limiters for each databases being searched. These are listed below the basic workspace where they normally appear. This allows you to set pre-search limits which are tailored to each individual database. In the example below you can see the pre-search limiters available for each database when you search RILM and MLA simultaneously.
The ultimate multi-database search engine at CWRU is called Summon. It searches all the resources for which the library has subscriptions. This includes all our databases and electronic journals. If even searches our streaming audio database such as Naxos Music Library or Smithsonian Global Sound. The single search box in the upper right hand corner of the Kelvin Smith Library website is the simple search space for this tool. The link under it labeled Advanced takes you to its Advanced Work Space. You should notice right away that there is only one search box and there is no choice for index. It is a massive single box keyword search engine. As a searcher you must rely very heavily on pre-search and post-search limits to manage the large amount of data you will retrieve.
There is usually a small suite of tools which accompany a given database and are intended to provide the user with online help, citation generation or management and saving or sharing data.
When using help screens it is important to differeniate between help screens for a specific database (covering content or indexes available) and help screens for the interface (how different search functionalities operate). Some databases/interfaces provide contextual help -- the help button retrieves help related to the page you are viewing.
Many databases provide citation generators which show how a citation for a specific database record would look using different stylesheet guidelines.
The possible means by which a user can now capture data from searches -- citations, full text, search strategies, or persistent URLs which lead back to a search or a citation -- has increased in the last few years. One principle of good searching applies to them all: save your work often. Most interfaces/databases have time-outs and these are often triggered by inactivity, so don’t leave your computer for any length of time without capturing your work in some manner. There are three aspects to capturing data (in this order): point of capture, mode of capture and formatting of data. Typically you tag a record in some manner, either from a results display or while viewing a full record.
Saving records to a file on your computer (or a flash drive if you are using a library computer), printing records, emailing records and exporting records are available in most interfaces. Exporting records often allows the added formatting option of choosing a citation style.
Programs like RefWorks are called bibliographic management software. They allow you to create and store whole libraries of citations. Within this software you can manage the citations by arranging them in different folders and generating bibliographies in different citation styles. Databases which allow direct export to these programs (as do the EBSCO databases) work the best. Data from other databases often requires some manual editing once they have been exported. CWRU has purchased and made available RefWorks free to all of its students and faculty. You only need to create a user account. There is an excellent online tutorial which goes with it, but it is a sophisticated piece of software and requires an investment of time in order to make it work for you. If you are writing a 10 page paper it may not be worth your time. If you are writing a doctoral dissertation, it may well be worth the effort. Keep in mind that difficult citation problems may still require you to understand the specific citation rules and to use judgment in how they are applied.
Recent developments in data capture include exporting links for searches and citations to online shared bookmark sites such as del.icio.us. This could be useful for research projects involving a group of people (e.g. the four members of a string quartet looking for articles which contain analyses of a given quartet). Searches may also be saved as RRS feeds which would allow you to reexecute the search at any time to see if new citations have been added. This could be useful for long-term research projects such as a doctoral dissertation where periodic checking for new articles on aspects of your topic is necessary.
Note that some interfaces allow you to set up user accounts where you can store citations or searches. If you do so, be sure to back up your saved data some other way. Some of our database subscriptions are based on the number of simultaneous users and once in awhile you may not be able to access the database or the data you saved in your user accounts.