Emulation has become an important issue for librarians and archivists seeking to store and provide access to computer systems for future generations of technology researchers. It is important to remember that also there exists a social and human context within which this software is used. This is especially visible with regard to video games. Not only does the text of a video game have the potential to be quite different each time the player interacts with it, but the experience of playing a video game can also often be quite social one. It is therefore beneficial to try to provide some account of the social and cultural context in which these games were played. The Gamer Feedback digital library project is an attempt to digitize for preservation this context, through recording oral narratives of ordinary people who played video games over the past two decades.

Creating a collection as a digital library allows for the possibility of providing access to the collection on the Internet. Additionally, with the right system implemented, there is the potential for gamers from other countries to submit their own audio narratives electronically to administrators to be included in the digital library.

More can be read about the genesis of the project on the initial project page which I created to discuss the project, people and terminology involved. You can also take a look at images of the project and of Greenstone on a flickr set I’ve created.
I worked on this project with a fellow MLIS Student, Rebecca Hale, as an assignment for our course on Digital Library Systems. Our assignment was to build a small demo collection (8-12 library objects) using Greenstone. This post is a slightly altered version of the short de-breif report I submitted as part of the course assessment.

Initial Questions

We began our project by examining the above-mentioned motivations for creating the digital library, as this would affect the way we would go about creating the objects for the project. I imagined our main clientèle to be a mix of game studies researchers and other interested parties wanting to explore the change in attitudes towards gaming over time, and interested in the analysis of players’ own articulations of their engagement with video game texts. For many social science and humanities researchers, how audiences make sense of and describe media texts is just as important as what the authors of those texts had in mind when they created them. As we were creating the interviews ourselves we spent some time deciding on questions to ask the individuals we would interview to ensure that they responded with useful content for the project.We also needed to decide what kind of files we should create. We knew that we wanted to record audio, but we also felt that it would be useful to the library user if we also included a transcript of the interviews. I also believed that it should be possible for the library user to read the transcript and listen to the audio at the same time.

After some experimentation with Greenstone, I discovered that it was possible to embed an mp3 file in an html page. I discovered that Greenstone only provided plugins for the mp3 format. We decided that we should include two files for each object in the database. An mp3 audio file, and an html transcript file which included an embedded link to the related mp3. In this way the user could choose whether or not they wished to listen to just the audio file, or see the transcript as they listened to the audio recording.

While Greenstone required us to use mp3 format for our audio files, this would most likely have been our choice at any case, as mp3 is a universally accepted format, especially in an online environment. As mp3 is so widely used, it is almost certain that it will be possible to convert from mp3 format into any future format that becomes popular.

Recording the audio files:

For my own interviews, I recorded straight onto a computer wherever I was interviewing. Initially, I used an podcasting website called Odeo, and used their flash-based online recording tool to save and store my interviews. This did work, and perhaps produced the best audio quality out of all of the interviews, however I found it a bit unreliable, and had to repeat one of my interviews because it was not saved to the site correctly. Later on, I decided to use a portable version of the well-known podcasting application Audacity, which I carried around on a USB pen drive, allowing me to carry my software with me, and store my recording on a portable medium.

Preparing the interviews:

I used Audacity again to crop any extra blank track from the beginning and end of the recordings. I also adjusted the volume in some cases. When Rebbecca’s audio files arrived they were in .wav format, because she used a different recording method to me, so I converted them to mp3. I named all the files in the same way using the first four letters of a Narrator’s name followed by the first four letters of the game they were talking about. I signified audio files with a capital “A” at the end of the file name.

As we were finalizing the audio files for the collection, we also worked on digitizing our respective interviews. Once the interviews had been digitized, I transformed the text files into simple html documents for inclusion in the collection. It tried to use as little mark-up as possible in order to leave the text in plain, readable ASCII format. This should mean that the transcripts can be easily converted from html into other formats, if necessary. I used the same naming convention with the transcript files, however I signified transcripts with a capital “T” at the end of the file name.

At the beginning of each html transcript I included the html code to embed the mp3 audio file at the top of the html page. I would have preferred to have used a more sophisticated audio player interface, perhaps something built in flash – however I found I did not have the time to create my own, and open source flash players I found on the web would not work correctly with Greenstone. Additionally, Greenstone’s default settings of blocking associated files in html documents would have meant that for each transcript page any related JavaScript code for a player interface would have needed to have been included in full in the head of the file. This would have made the html document excessively cluttered and messy. I do think that, if I had just spent more time exploring the way that Greenstone operates, I could have cracked this issue. As it stands, I felt it was best to stick with the much cruder, yet simple, embed tag to create the effect at this stage. The most important drawback of making this choice is the fact that the user is forced to run the library in Internet Explorer if they wish to hear the entirety of the audio files. Additionally, as I understand it, the embed tag is not standards compliant code. If I were taking this library further, I would want to resolve this outstanding issue.

Creating the library:

Our decision to make each library object consist of two complimentary files greatly influenced the way the collection was structured. I assigned each library object a folder of its own and placed the two components of the object within that folder. In this way, in applying the metadata, I was able to take advantage of the inheriting feature of Greenstone which allowed files further down a hierarchy of folders to inherit metadata values from parent and grandparent folders.Using the main ‘library’ folder as my starting point, I was able to easily assign collection-wide metadata values such as dc.rights and dc.publisher to all of the files in the library. I was able to assign much of the metadata values at the folder level for each library object. I was only necessary to input metadata at the individual file level for values such as dc.format and dc.relation and this made it much faster to apply metadata to the collection.

This hierarchical approach to organizing the files also had useful repercussions when designing the user interface. In the “title list” view, Greenstone was able to easily display a list of all the objects in the collection based on their folders, and then provide an expanded view of the particular files within each of those objects.

I changed some of the metadata that had been provided to me by Rebecca in order to make full use of the Greenstone’s ability to organize metadata terms hierarchically as well. This was especially useful when providing descriptive metadata about video game genres mentioned in the interviews.

Once more, this proved useful in the design of the digital library interface, because it allowed me to create a keyword hierarchy that sorted the files by keywords used to describe them. I was able to create both a full alphabetical list of all the metadata used, but also a hierarchical list that allows users to browse the keywords by their relation to one another.

I also used the applied metadata to partition the collection so that users could search by a particular game genre as well as by a particular field. I believe this will be very useful as the collection gets larger and contains discussions of a greater number of games that might fit into these genre groupings. It is interesting to note that, with our current sample, you can often return results when searching for something that does not fit into the genre category you have selected (for example select ‘Genre: Versus Fighting’ and search ‘Full Text’ for “Sonic” – which is actually a Platformer game). This is because we asked Narrators to briefly talk about games they played in addition to their game of choice, resulting in them mentioning a number of titles early on in their transcript.

I would be interested in exploring ways that the metadata could indicate which descriptive keywords referred to games that the narrator was talking about in detail, and which were just passing references to other titles played at one time. Alternatively, perhaps it would have been better not to ask Narrators to reflect in this way at the beginning of their talk at all. A savvy user can still find library objects that deal mainly with a certain game if they search by Title, due to the fact that the naming convention of each library object includes the name of the game that the person is speaking the most about. It would have also been useful to include the exact date that each game was published as part of the descriptive hierarchy of terms associated with it.

Concluding thoughts:

While I think our project produced a well-described and navigable library, I nonetheless feel that I must have only scratched the surface of Greenstone’s potential as a Digital Library Management System. For example, in this digital library there needs to be more flexibility for the user to browse between library objects based on subject keywords and other metadata terms. Presently, the user must always enter from the top - via search or browsing - and then return to the top level once more if they do not find what they are looking for or wish to start a new search. I experimented with implementing several techniques I have seen in other digital libraries and library catalogues, but was unable to produce a satisfactory result. I feel somewhat hampered by my lack of programming knowledge, but at the same time feel that - had I simply been building this as an html or php website (with more freedom as to where I could put links and site elements) I could have done much more to create the effects I would want to implement.

I think we could have done better in the metadata to describe more exact dates of games and dates mentioned by Narrators. I didn’t discover Greenstone’s ability to create hierarchies of metadata terms till quite late in the project - after Rebbecca had finished defining the metadata - and I think that knowing about this earlier on could have enabled us to make the descriptive metadata a bit clearer.

Nonetheless, I am quite pleased with our project - and would like to think about continuing it as some kind of online project. I think, however, that this would might require some kind of human ethics approval for it to exist beyond simply as a student project.