Baron / TXWI-C / p. 1
Benjamin Baron
Spring 2009
SLIS 5200 TXWI-C
Final Draft 4
The North Texas Rock Music Appreciation Society:
Information Organization System
1. Project description
1.1. Collection and information objects
The North Texas Rock Music Appreciation Society collection consists of 1276 compact discs, housed in a converted Victorian style house in downtown Frisco, Texas. These musical works span a variety of Rock and Roll genres, from both American and European artists, and all objects in the collection were produced after 1960. Donations and Society purchases are the primary method by which the collection is developed.
The Society exists for the purpose of educating young people in music appreciation and in exposing members to styles of music to which they are not otherwise exposed. This collection serves many of the suburban areas north of Dallas, Texas.
1.2. Users' demographics and knowledge
Membership in the Society consists of 337 official members. Patrons of the collection are drawn from a variety of backgrounds, and represent all racial groups within the local community, with the majority of users coming from Caucasian and Asian backgrounds. The ages of members range from 16 to 37 years, representing a wide cross-section of experiences and socioeconomic backgrounds. Overall membership is represented by large groups of secondary school students and college students, with a much smaller percentage of postgraduate students existing within the group.
A wide variety of life experiences are represented, especially when considered in terms of age and education. Domain knowledge represents collected knowledge on a subject or field of inquiry. Members of the Society are often self-educated in the field of rock music, exhibiting a moderate level of domain knowledge on the subject of rock and roll. In terms of the ability to use computer based search systems, members show a moderate level of system knowledge, as they are familiar with the use of computers and search software to find and access information.
Finally, members display high information seeking behavior, as they often actively seek out new information connected to information they already possess. This group of users shows a high level of sophistication as evidenced by the sometimes complex and specific nature of their questions. This level of behavior results in users consulting a variety of resources regarding the subject in which they are interested, such as magazines, websites, video resources, and other text resources. Such behavior also results in the use of narrowly scoped initial information in order to seek out related materials of a much wider scope. Based on these user characteristics, an information system for this collection must be able to address moderately specific questions while presenting the user with a variety of search options.
1.3. Users' problems and questions
Member inquiries often address a variety of terms describing the material within the collection. Among the topics of these questions are artists or performers, producer information, copyright dates, and related sub-genres. All of these questions can be primarily addressed from the subject of musical sub genre, within the overall genre of rock and roll. Members of the society often seek to expand their musical horizons and knowledge by researching musical acts or recordings related to their current level of domain knowledge. Often this is accomplished by searching for information related to artists within a genre or by delving further into related genres. Members may also seek information regarding more esoteric aspects of a musical work, such as names of producers, nationalities of artists, or the presence of mutual members of a musical act.
Member questions exhibit wide variation in both the precision and desired recall of their searches. Precision refers to desired specificity of actual search results to a search query, and recall is a quantitative expression of how much information a user desires in order to address their question. Based on the users’ levels of knowledge as described, the following represent typical questions regarding the collection.
User question 1: “I am preparing a presentation on the Post-punk movement, and I need around seven to ten examples of musical works that developed out of this movement.”
Object attributes: Genre.
Desired precision: Low. The works are not asked for specifically, rather a range is necessary.
Desired recall: High.
User question 2: “I enjoy Industrial music, especially the album “The Downward Spiral” by Nine Inch Nails. Can you show me 3 or 4 more titles involving the same band members or producers?”
Object attributes: Genre, album, producer, artist, band members.
Desired precision: High. Not only is it necessary to identify the members and producers of the album “The Downward Spiral”, but it is also necessary to then find titles involving those same individuals.
Desired recall: Moderate.
User question 3: “I’d like one or two psychedelic albums performed by British musicians after the 1970’s.”
Object attributes: Genre, nationality of artist, copyright date.
Desired precision: High. Not only does the user specify the nationality of the artist, but albums released after a specific date are required.
Desired recall: Low.
User question 4: “My parents don’t want me to listen to music with explicit lyrics. Can you show me some metal albums that I can listen to without getting in trouble?”
Object attributes: Genre, content warning.
Desired precision: Moderate. The user wants a variety of objects to choose from, but those objects must exhibit a specific characteristic.
Desired recall: Moderate.
The user questions address a reasonable scope of attributes describing the materials within the collection. Based on the questions presented, it is necessary to include the attributes: genre, sub genre, album, producer, artist (including individuals and bands), nationality of artist, content warning, band members, copyright date.
Additionally, and for the purpose of including sufficient data for each object in the collection, a number of other attributes can apply to the objects in this collection, namely: song titles, label (publishing studio), special features (such as multiple discs, or SACD format), and UPC.
2. Representation of information objects
2.1. Entity level
Entity level addresses the physical form and scope of any one object in a collection. The simplest example of an entity is a single piece of media such as a compact disc. Compact discs can be considered either as whole entities or as a collection of entities, or individual tracks of audio in this case. The manner in which a collection’s entity level is determined is based on the needs of the collection’s users.
Entity level is the initial basis upon which collections are organized and sets the method by which users may search a collection. This is an important distinction that must be made at the beginning of collection development, as decisions made at this step of the process affect the organization and usability of the overall collection. For example, it does not make sense to organize and classify a collection of music around recordings of two thousand individual audio tracks. While such a system presents very precise results to users, it is not practical for collection organization and causes significant changes to the search behavior required to browse the collection. A better choice in this case is to organize the collection around an entity level consisting of individual compact discs, which fulfills the users’ needs while creating a manageable organizational system.
For the purpose of this project, the entity level of the collection is based on single album titles of audio compact recordings; therefore, within the system, one database record represents one compact disc recording. Compact disc recordings that consist of multiple physical discs are considered as a single whole, based on the title of the album. This entity level is appropriate for the North Texas Rock Music Appreciation Society collection, as the entirety of the collection is built around the physical format of compact discs. This entity level is also appropriate for the users of the collection, as the majority of user search behavior is conducted based on a whole compact disc level.
2.2. Metadata elements and semantics
Metadata schemes organize and describe the attributes of an entity. Metadata elements within a metadata scheme represent the attributes of an entity as fields within a record. For example, a metadata scheme for a collection of audio recordings contains metadata elements including creator, title, publication date, genre, UPC, and classification, among many other potential metadata elements. The creation of a metadata scheme allows the database designer to organize a collection based on similar elements of individual metadata, eventually converting these elements into fields within a database file.
The North Texas Rock Music Appreciation Society’s collection is organized around nine metadata elements. A number of additional attributes are addressed in section 1.3, including song titles, label or publishing studio, special features, and UPC. For the purpose of this project, the attributes of song titles and record label have been eliminated as essential attributes for inclusion as metadata elements, as they do not address the user questions addressed in section 1.3.
Metadata addresses four specific user tasks: finding, identifying, selecting, and obtaining. In the task of finding a user applies known information, or conducts a search of the file, in order to find records that match the search criteria. For example, if a user knows the name of an artist, the user can then use that information as a search term to find related records. Elements that address the user task of finding are Creator, Title, and Subject, as these elements represent the most common attributes that a user knows about the object it is attempting to acquire. For example, a user searches for the Creator name “The Beatles”, and is presented with the names of several albums. The user has accomplished the task of finding by utilizing the Creator element.
Identifying is the process in which the user determines which record or records most appropriately correspond with the original goal of the search. For example, a user searches based on known information, such as Creator name, and then uses other information to identify which records are more closely related to the search criteria. In the metadata scheme for the North Texas Rock Music Appreciation Society’s collection, elements that address the user task of identifying are Creator, Title, Date, Subject, Feature, Nationality, Producer, and Identifier. All of these metadata elements relate to fields within the record that allow the user to identify specific features about a record that assist in identifying which records are appropriate to the search goals. Using the results from searching “The Beatles”, the user now identifies which album or albums meet the search criteria. In this case, the user is looking for the titles of albums produced before 1966. By utilizing the Date element, the user has identified a specific object or specific objects within the search results.
Selecting occurs after records are found, and after specific records are identified as appropriate to the search goals. Selecting is a process in which the user selects records from the file that most closely match the search criteria. At this point in the process the user may alter the criteria upon which they are searching based on selections related to the search results. As with identifying, a greater level of specificity is sought, typically by using the information contained in the returned records. Elements within the metadata scheme that address the user task of selecting are Creator, Title, Date, Feature, Nationality, Producer, and Identifier. The inclusion of these elements provides users with detailed information contained in the records, and assists in the user task of selecting by allowing users to discriminate between the records. Now that the user has found and identified results using “The Beatles” and production dates before 1966, the user selects the most appropriate record or records by utilizing further information. If the user is searching for a re-mastered recording of the album, a record that fulfills this desire is selected using the Feature element. Since the North Texas Rock Music Appreciation Society serves a user group of varying domain knowledge the inclusion of specific fields serves those users who require narrowly focused information. While the most basic elements such as Creator, Title and Subject serve all users, more knowledgeable users require the use of elements such as Date, Feature, Nationality, Producer, and Identifier to produce results appropriate to sophisticated search criteria.
Finally, the task of obtaining occurs when a user gains access to the object or acquires the classification information that indicates the physical location of the object within a collection. The Classification element directly addresses this user task by providing a location address to the user. In the example, the user finds, identifies, and selects a specific version of an album produced by the Beatles previous to 1966. Using the Classification information, the user can now physically acquire the results of the search.
Metadata elements within a metadata scheme can be described as container or content based. In the case of a container based element, such as Title, Date, and Identifier, information regarding the physical object is conveyed, while for a content based element, such a Subject or Classification, a relationship within the collection is described.
A complete list of the metadata elements and their related semantics are found in Appendix A, Metadata elements and semantics.
2.3. Record structure and specifications
In the creation of a database it is necessary to create fields that are connected to the metadata elements within the metadata structure. Metadata elements are sometimes represented on a one-for-one basis by the fields within a record, but in some cases it is necessary to split a single metadata element into multiple related fields. One instance of such a split is seen with the Creator element. Cases exist where multiple entities serve the role of creator for a musical work, such as composers, musicians, and arrangers. While all of these entities are creators, they each serve different roles in the creative process, and it may be appropriate to distinguish their individual roles within the database by defining specific fields, even though the fields all stem from the same initial metadata element.
There are fourteen fields present in the database for the North Texas Rock Music Appreciation Society’s collection. In the structure of the collection’s database there are two automatically generated fields: RecordID and RecordDate. RecordID refers to the unique database identifier for a record. This identifier is generated at the time of record entry. Therefore, the tenth record entered into the database is automatically assigned the RecordID of “10”. RecordDate refers to the actual date on which the record within the database was created. RecordID is indexed within the database, however RecordDate is not indexed. While neither field addresses the four user tasks in an efficient manner, RecordID has the potential to be of value to technical users, and so is automatically term indexed. Entry validation is automatically applied by the database software, and content validation is not appropriate as each entry is unique to the record as it is created.
In this project there are three instances of metadata elements that are split into multiple fields. First, the metadata element of Creator is split into the fields of Artist and Member. Artist refers to the individual artist or the musical act responsible for an original audio recording. Member refers to musicians who are members of a musical act. It may also refer to the individual artist referenced in the Artist field.
Second, the metadata element of Feature is split into the fields of Special Features and Content Warning. The Special Features field refers to qualitative features of a recording, such as Super Audio Compact Disc encoding, gold recording media, or a title that consists of multiple compact discs. The Content Warning field indicates the presence or lack of a content warning for an audio recording.
Finally, the metadata element of Subject is split into the fields of Genre and Sub Genre. Genre refers to top level terms which describe the music contained on each audio recording. Sub Genre refers to stylistic distinctions within a genre, and contains several levels of progressively specific yet related terms.
Field type addresses the technical format of a field. Common field types are text, number, and date. Within The North Texas Rock Music Appreciation Society’s database, five field types are utilized. The majority of the fields within the database utilize a text based field type, specifically the fields of Artist, Member, Title, Genre, Sub Genre, Special Features, Content Warning, Artist Nationality, and Producer. Entries in the Classification field utilize an assortment of alphanumeric characters; therefore the use of the text field type is appropriate for this field. The two exceptional fields within the database are the Date field, utilizing a date field type, and UPC, utilizing a number field type, as is appropriate to the respective content of these fields. Finally, RecordID utilizes the field type of autonumber while RecordDate utilizes the field type of autodate, both of which are specialized field types automatically generated by the database software.