The Beer Snoob Thesaurus
  Home   About  User Guide Analysis Thesaurus Resources Search Disclaimer


Structural Analysis of the B.S. Thesaurus



Suitability for the Environment
 
At present, the Beer Snoob Thesaurus contains terms in five primary areas of interest to users: appearance, flavours and aromas, ingredients, mouthfeel and styles. Over half of the terms in the initial version of the thesaurus relate to beer styles because they are more accessible to fledgling snobs than the less tangible terms related to flavours and aromas and mouthfeel. As well, most Beer Snoobs tend to search for beer reviews using style.  In later versions of the thesaurus, terms will be added, including many from the difficult-to-describe categories flavours and aromas and mouthfeel. Essentially, the thesaurus itself will develop in response to the needs of its increasingly sophisticated (and, of course, snobbish) users.
 
All members may assist with this project by contributing reviews and entering basic data into the database. Currently, all indexing is done by trained librarians who are also committed snoobs. As mentioned in the About section, in the future some members may be taught how to index by Beer Snoob librarians or, more likely, indexing internships will be offered to local library school students.
 
return to top
 

Type of Indexing Language
 
The Beer Snoob Thesaurus is a controlled vocabulary constructed after reviewing the terminology and indexing languages used by sources on the subject. It follows the ANSI/NISO Z39.19-2005 Guidelines for the Construction, Format, and Management of Monolingual Controlled Vocabularies. (Note: Parenthetical references appearing below direct the reader to a specific section in these guidelines.) As such, the thesaurus does not make use of natural language. However, users will be able to search the online database using natural language in the form of keywords.  
 
Each single-word and multi-word term in the thesaurus represents a single concept (6.3.1). When a single concept could be represented by more than one term, a preferred term was selected from among these terms based on literary warrant (8.2; 6.6.1.1). 

return to top
 
 
Post-coordinate Retrieval
 
The thesaurus has been designed to facilitate post-coordinate retrieval, which allows users to employ “Boolean and other operators to express complex concepts” (7.2.2.1). The online database will allow for post-coordinate retrieval. For example, users may search for reviews of creamy stouts by combining the following indexed terms: creaminess AND stouts.
 

return to top

 
Capitalization
 
In the literature, some terms were found equally as often with initial capitalization as they were entirely in lowercase. Doppel bock, for instance, appears in the Oxford English Dictionary as “Doppel bock” or “doppel bock.” In such cases, the thesaurus uses the lowercase form, following ANSI/NISO (2005) recommendation 6.7.1, which states that “predominantly lowercase characters [should] be used for terms in controlled vocabularies.” In agreement with recommendation 9.2.5, the thesaurus serves as an orthographic authority and thus does not uniformly use lowercase initial letters. Where merited by the literature, initial capitalization is used.
 

return to top
 


Pluralization
 
Some nouns in the controlled vocabulary follow the ANSI/NISO (2005) recommendations regarding nouns in a straightforward way: count nouns such as “oats” appear in the plural (6.5.1); mass nouns such as “foam” appear in the singular (6.5.2). However, the nouns describing beer styles present a more complicated situation, partly due to the inevitable addition to the thesaurus of more terms. Here the creators of the thesaurus follow the ANSI/NISO (2005) recommendation that, given user-warrant, a class with more than one member be expressed as a plural (6.5.2). Thus “stouts,” which beer snobs regularly break down into subtypes such as “oatmeal stout,” is expressed as a plural. “Oatmeal stout,” on the other hand, does not have related, snob-approved sub-types and so is expressed as a singular noun. Nouns describing style were analyzed in this way: even if a term for a beer style is the narrowest in a hierarchy as the thesaurus currently describes it, it may appear in the plural because the thesaurus could conceivably be expanded to include sub-types of that beer style, if those styles later showed user warrant. One example of this latter scenario is “Scotch ales,” which appears in the plural in case the thesaurus ever needs to be revised for more specificity in this area.
 

return to top
 

Grammatical Forms of Terms
 
As suggested by ANSI/NISO (2005), terms throughout the thesaurus appear in the form of nouns and noun phrases (6.4.1). For commonly used terms in grammatical forms other than nouns or noun phrases (e.g., acidic and cloudy) these terms were included in the thesaurus as non-preferred terms; these terms direct users to preferred terms (e.g., acidity and cloudiness). As the thesaurus develops, more non-nouns and non-noun phrases (in particular, more adjectives related to appearance, flavours and aromas, and mouthfeel) will be added as non-preferred terms in order to direct users to preferred terms.
 

return to top

 
Relationship Structures
 
The thesaurus denotes three types of semantic relationships: equivalence, hierarchical, and associative. 
   
Equivalence Relationships
 
Two or more terms expressing the same concept have an equivalence relationship (8.2). In such instances, one of these terms has been chosen as the preferred term, with equivalent terms appearing in the thesaurus as non-preferred terms (8.2). Users encountering a non-preferred term will be directed to its preferred term (e.g., collar USE: head). As previously mentioned, the preferred terms were selected based on literary warrant (6.6.1.1). Common equivalence relationships include synonymy (e.g., special bitter / best bitter) and lexical variation (e.g., Pilsners / Pilseners). Over time, more equivalent terms will be included in the thesaurus as non-preferred terms due to an abundance of extant synonyms and lexical variants.  
 
 
Hierarchical Relationships
 
Hierarchical relationships indicate superordinate and subordinate connections between terms. A superordinate (parent) term is a broader term, namely a class or a whole, and subordinate (child) terms are narrower terms, which represent the members or parts of that class or whole (8.3). For example, cream ales (NT) are types of lager ales (BT). 
 

Associative Relationships

Terms with an associative relationship lack equivalence or hierarchical relationships but are semantically or conceptually related (8.4). Related terms (RTs) were determined on the basis of their value to users and/or indexers. For example, if the extent of the relationship between two terms is simply that both are ingredients (e.g., water and wheat), this relationship would be of little value to users and indexers; therefore, these terms would not be represented as having an associative relationship. On the other hand, it is often valuable to identify relationships between terms which would otherwise remain concealed, such as the relationship between hops (an ingredient) and hoppiness (a flavour/aroma). These terms are clearly related; however, they appear in different hierarchies, meaning that the only relationship they can be given is associative. Likewise, the relationship between antonyms (e.g., thinness / fullness) may be usefully represented as associative. As a result, users looking up thinness in the thesaurus will be reminded that it is related to fullness. In this thesaurus, associative relationships have been identified when they were deemed valuable for users and/or indexers.
 

return to top

 
Precision
 
Precision, the number of relevant items retrieved out of the total number of documents retrieved, is served in several ways. First, the thesaurus is designed primarily for post-coordinate retrieval. Post-coordination will allow users to narrow their searches through the addition of terms using the Boolean operator “AND.” Secondly, the thesaurus employs scope notes to clarify the prescribed meaning of terms within the indexing language. Assuming indexers conform to this vocabulary control when indexing the club’s review collection, the collection will be searchable with a high degree of precision. Thirdly, the thesaurus carefully constructs relationships between terms according to the warrant of a broad survey of the literature. Hierarchical relationships in particular enhance precision by leading searchers to narrower terms. The thesaurus authors’ careful and economical identification of associative relationships also increases precision by suggesting overlapping terms while at the same time not leading users to unduly broad searches. Finally, the specificity of the language (which should increase as terms are added in the future) guarantees high-precision searching.
 

return to top

 
Recall
 
Recall, the number of relevant documents retrieved out of the total number of relevant documents in the collection, is also enhanced in a few different ways. First, control of word forms such as singular and plural reduces the chances that items will be missed due to noun form (such as singular and plural). Secondly, control of synonyms and quasi-synonyms, such as lexical equivalences and European vs. North American terms, reduces the chances of missed documents. Thirdly, structural syndetic relationships strengthen recall by leading searchers to broader terms and by suggesting related terms that could be used to expand a search using the Boolean operator “OR.” Fourthly, and more specifically to the domain of beer reviewing, recall is improved by the combination of terms for flavour and aroma into one hierarchy. Because these terms are generally the same, and the two aspects of beer are almost always discussed together, it made sense to combine them in this way, thus slightly increasing recall.
 

return to top

 
Specificity
 
The specificity of the Beer Snoob indexing language—its ability to point to the details of a beer review—was given careful consideration by the authors. The thesaurus aims to provide a controlled vocabulary at a level of specificity suitable to the club. In this initial version, the thesaurus attains a high level of specificity in dealing with the major styles of beers discussed in North American beer literature. It also provides a few key terms related to ingredients as well as the less tangible areas of mouthfeel, appearance, and flavours and aromas.
 
Further care for the appropriate level of specificity is evident is in the choice of node labels. To enhance the clarity of certain hierarchies, non-indexable terms describe the top of the hierarchy. For example, “flavours and aromas” is a node label, rather than an indexing term, because practically all beer reviews discuss the flavour and aroma of beers, but the term is useful for showing the principle of organization in the hierarchy. “Mouthfeel,” on the other hand, while commonly discussed, is by no means a required consideration in every beer review; it therefore provides a useful, if broad, indexing term, not just an indicator of hierarchical structure.
 

return to top

 
Exhaustivity
 
The most exhaustive aspect of the thesaurus is that dealing with beer styles, which stops short of brand names and individual labels. On the other hand, this vocabulary is not wholly exhaustive, since in some cases more specific sub-categories of beers may conceivably appear (either as types discussed more extensively than at present—providing literary warrant for their inclusion in the thesaurus—or as new types entirely). For further discussion, see the section on singular and plural noun forms.
 
Terminology for ingredients, appearance, mouthfeel, and flavours and aromas represents the most basic terminology used in beer-reviewing. Users can expect this vocabulary to be further developed in future iterations of the thesaurus.
 
Additional classes of terms may also appear in future versions: for example fermentation processes, bottling, and drinking occasions. These terms have been excluded in the present version due to its scope as a first foray into organizing the Beer Snoob review collection.
 

return to top

 
Relative Weight of Aboutness and Meaning
 
At present, The B.S. Thesaurus is primarily a controlled vocabulary for describing the aboutness of beer reviews in the B.S. Collection. The thesaurus has the most depth in describing and relating reviews in terms of the beer styles under review. With a search and retrieval system, the controlled vocabulary will allow snoobs to collocate reviews of beers based on the objective criteria of style, even if the reviews themselves vary slightly in spelling or usage of these terms. Terms for flavour and aroma, as well as appearance, are also generally aboutness terms; however, as the collection grows and the indexing project goes further, this and other areas of the thesaurus should be developed to accommodate more personalized searching objectives. For the time being, however, through its emphasis on aboutness, the thesaurus is helpful to aspiring beer connoisseurs who will most likely want to begin reading up on recently reviewed beers according to style, thus allowing them to order or buy these beers with a minimum of uncertainty and embarrassment.

return to top




  
LIBR 512: Indexing        Submitted March 15, 2007        J. Haigh, M. Holle, S. Plante, C. Watt
LIBR 512: Indexing        Submitted March 15, 2007        J. Haigh, M. Holle, S. Plante, C. Watt