Jessica Browning
Assignment #2: Final
December 10, 2004

One problem with consistency in metadata creation was inspired by the DLESE Quality EG 3 Metadata Structures white paper made available as a lecture note.  In the white paper, the work group discusses the problems dealing with "standardizing" metadata vocabularies to facilitate information retrieval.  While the white paper only discusses information retrieval in terms of assigning subject terms to metadata, I believe that a review of classification systems--in particular, the Library of Congress cataloging system, which I used-- can also be beneficial at this time, due to the research being done on metadata vocabulary standardization.  When I researched my selected topic ("Japanese kanji online resources") using Google, I encountered a variety of sites, not all of them immediately relevant.  Initially, I blamed my search skills; however, a quick visit to the Library of Congress's classification scheme revealed that every part of the Japanese language--from kanji to classical Japanese to local dialect--is lumped under PL501-699: Japanese Language.  To me, this seems to be a little dismissive of the fact that Japanese is a complex language made up of many elements.  One can argue that a broad classification is more than adequate to the casual researcher--after all, perhaps not everyone realizes that Japanese has many elements--, but to the more advanced user (a person who knows "exactly what they need")  a broad classification could slow the retrieval process because the user has to pick through several resources to find what he/she needs.  

In my initial report, I rather boldly stated that "If I were actually cataloging the resources for a library, I would perhaps try to further subclassify the "Japanese language" into smaller categories such as "Japanese kanji" and "Japanese grammar".  However, after taking a bit of time away from my assignment, I now realize that this may not be the brightest idea I've ever had.  After all, if a user is expecting to find resources under "Japanese language", and I've divided resources into my own arbitrary categories, then a patron will have a hard time finding anything.  To counter this problem, I listed "Japanese language" as a subject term, and then created my own terms based upon what I "thought" should work.  Then, within my metadata, I tried to use the same arbitrarily created terms in order to make my metadata consistent.  I was mostly able to stay within the same terms for all 10 resources.  However, I still ended up with a couple of unique subject terms such as "Japanese alphabet system" for Resource #9: CMJ Grammar Online and "Japanese kanji furigana" for Resource #10: Kanji Furigana For Japanese Learners.  Even though the subject terms accurately reflect the information available in the web site, I'm not sure that a patron would think of those specific search terms to find those resources.  To completely cover all possible subject terms, I would give Resource #9 two more subject terms: "Japanese hiragana" and "Japanese katakana".  Then, keeping in mind that a beginning Japanese student may not yet know/remember the word "furigana", I would give Resource #10 another subject term called "Japanese kanji readings". 

For a controlled vocabulary point, I prefaced every subject listing with “Japanese”.  This was a good move in terms of making distinction between Japanese kanji and Chinese kanji, but it also makes sense in terms of being a cataloger.  According to the metadata creation text, my use of "Japanese" also leads to improved subject information retrieval Using the same terms to describe a group of resources was relatively easy—I just had to scroll up to see what I had entered before.  However, what would happen if I wouldn’t have cataloged everything at once?  I would have to first have a good knowledge of what was available in my library, and then be able to find the appropriate record or records.  By prefacing everything with “Japanese”, it doesn’t matter so much if I can remember whether I said “Japanese kanji” or “Japanese writing system”—I only have to search for “Japanese language” or “Japanese”, and my resources will hit.  In addition, I was also careful to use abbreviations as well as full text—JLPT and Japanese Language Placement Test.  While the actual name of the test is “nihongo nouryoku shiken”, most English speakers taking the test probably realize that materials are classified under the English translation, or else using roman characters to search for Japanese names doesn’t always work so well.

2.  Accuracy
In creating my metadata resources, I followed the instructions provided in the cataloging text.  According to our instructions, we should "Take title from the actual information resource" (24).  I do, however, question a few of the titles given to the web pages.  For instance, as a researcher, I've been taught that in doing research, one should only use resources that "sound" credible.  Therefore, a resource titled NASA Guide to Locating Venus In The Night Sky is a credible source since NASA is an official space agency, but "Bubba's Guide to Finding Venus Among All Them Sparkly Dots" probably isn't so credible.   Even when I was locating resources to create metadata for, I unconsciously filtered out sites that didn't sound so good based upon the titles, the URLs, and the descriptions.  But, now that I look back at my metadata, I question whether, as a patron, I'd find the sources I chose to be very valid.  Most of my titles, such as "Kanji alive" and "Kanji a Day", sound okay, but I'm not so sure about "Jeffrey's Japanese-English Dictionary--Gateway" or "Charles Kelly's Online Japanese Language Study Materials".  In my first report, I had anticipated that the credibility of these sites might be questioned, so I justified my choices by explaining that "there aren't a lot of "really good" kanji sites available on the web--there are a few authoritative sources, and those are what other Japanese pages link to".  Ironically enough, in spite of their titles, Jeffrey's Japanese-English Dictionary and Charles Kelly's Japanese Study Material site are both well-known by Japanese language students. 

At the same time, the majority of my titles all have the word "kanji" in them.  This is an accurate description of the particular web site, but when I review my metadata, I have trouble remembering which web site corresponds to which title!  I can easily recall what Jeffrey's Japanese-English Dictionary and Charles Kelly's Online Japanese Language Study Materials contained, but I have to sometimes read the descriptions to remember the differences between "Kanji alive" and "Kanji-Step" and "The Kanji SITE". 

Now, with objectively analyzing my own work, there are a few things I want to question.   The biggest error I discovered was in Resource #9: CMJ Grammar Online.  The title itself suggests that the site contains primarily grammar.  Also, in my description, I only mention the grammar resources available.  Yet, in my subject terms, I list Japanese kanji.  Here, I have to admit my failure to be accurate.  If I were listing Japanese kanji as a search term, and the search "Japanese kanji" yielded this site, I should have told the patron how the site was relevant.   Another error I discovered was in Resource #3: Kanji-Step.  In my description, I stated that the user could see an introduction to all aspects of the Japanese language, and I also suggest that there are specific sections for hiragana and katakana.  However, in my subject terms, I only used Japanese language and JLPT--Japanese Language Placement Test.  To properly fix my record, I would include some more subject terms, such as "Japanese grammar", "Japanese hiragana", "Japanese katakana", and so forth.   Now that I reflect back on the time I was making the record, though, I remember that my initial focus was only Japanese kanji.  Therefore, I was only listing subject terms that seemed relevant to my chosen topic.   But, I now realize that if I were taking the time to create metadata for a resource, I should create data for the site in its entirety, and not only the parts relevant to my topic.  This oversight also occurred in Resource #4: The Kanji SITE, Resource #5: Charles Kelly's Online Japanese Language Study Materials, and Resource #7: Nihongo Web. 

3.  Granularity
In my research and metadata creation, I mostly presented web sites rather than individual web pages in my metadata.  While this may be questionable, a majority of my web sites featured information that was immediately accessible, if not one click away, from the home page.  Since I used Google to find relevant sources, and then documented the search terms I used as subject terms, I assumed that the entire content of a web site could be potentially useful to a patron.  In addition, though, even individual patrons may have separate reasons for searching "Japanese kanji".  For instance, one patron may input "Japanese kanji" to find the joyo (a government-approved list of 1,945 daily-use kanji) kanji list, whereas another patron may be looking for proficiency examination kanji lists, and another patron may want a definition and/or historical explanation of what kanji is.  In my own research interests, I'd rather be presented with web sites and then allowed to filter the useful from the non-useful rather than pointed to direct pages that immediately offer answers.  This idea, along with the idea that patrons may have different needs that stem from one topic, led me to present information from a "top-down" approach.  In my case, being a "top-down" researcher is okay, since I have the time to dig through web pages and sources to find what I want.  However, for busy people, being asked to find the proverbial needle in the haystack in 3 minutes or less might cause a bit of unnecessary stress.  As an information professional, whose job is to organize things so that they can be found easily, I've just failed in my job by asking people to "find it themselves".  This concept is discussed in a DLESE study titled "Merging Metadata and Content-Based Retrieval".  Using a conceptually-based hybrid educational resource discovery system that combines metadata and content-based retrieval methods to organize information, the authors conducted an experiment based upon two hypothetical situations: a teacher searching for a tool to use for one class, and a teacher searching for resources to use in teaching a 6-week module.  In presenting an overview on the design methodology and considerations that went into designing the system, the authors state that "many K-12 science educators are teaching out-of-area-... and thus lack confidence in their ability to evaluate the quality of science education resources" (3).  Lack of confidence means that "...considerable time can be required to comprehend a resource in order to determine if it is indeed relevant or not" (3).  As an information professional, I need to consider the needs of my entire campus community/patron base more than one group.  In creating my metadata, I focused more on students because of the subject I chose--I presumed that Japanese language professors were either going to already know kanji, or else they had their own resources for verifying a kanji.  This was a mistake, as a professor may need a tool to recommend to students and only have 10 minutes between classes to find the URL to use in the next lecture. 

But, at the same time, this assignment required that we locate and create metadata for 10 sources.  Yet, in reality, 10 sources for one subject is a little big for a casual researcher.  I discuss this in my first assignment's assessment when I state that  "In a realistic library setting, though, I don't believe that 10 links would be offered to patrons...if I were a patron faced with 10 different kanji sites to choose from, it would take a long time to explore and choose my resources."  However, a metadata creator may argue that more advanced students may be completing projects which require more sources.  In this case, though, I would suggest that a student writing, for instance, a thesis, wouldn't be using the Internet as his/her primary research source.  Also, in the case of my chosen subject, a more advanced student probably wouldn't be using the material I selected--the content is more referential than informative in nature.  While "subject needs" and "pedagogical needs" are named and outlined in our metadata handbook, two other factors which must be considered are the type of information institution being cataloged for and the needs of the overall patron base.  For instance, a university that doesn't offer Japanese as a potential course of study may only need 1 or 2 sources which define kanji, so a lot of time doesn't need to be spent on finding many useful resources when 1 or 2 will suffice.  Another instance would be a public library, where a patron needs a kanji dictionary but the library doesn't own a print version.  Therefore, the patron and the librarian navigate the Internet to find a good dictionary, and the librarian decides to simply bookmark the page in case it's ever needed again.  But, a university offering both undergraduate and graduate studies in the Japanese language would perhaps need more resources that cover a variety of kanji aspects (then again, Japanese language students will have other sources to verify kanji, such as electronic and paper kanji dictionaries--after all, I'm a Japanese student, and I use my electronic dictionary rather than the Internet to look up kanji).      

Part 2: Personal Criteria