New Tools for Film & Media Studies

A number of emerging technologies may soon revolutionize the analysis and teaching of film and media studies. The Motionbox site allows users to upload videos, highlight and tag individual scenes, search for similar tags in other videos, and share media with others through email or Web site posting.

Creating tagged video clips in Motionbox is as easy as highlighting a portion of the “filmstrip” and then adding keywords or phrases. Users can use the tagging system to create episodes (like DVD chapters) or label very short video clips. The site also allows users to “jump” to particular tagged sections.

Demonstration of the clip highlighting tool in Motionbox.

Image this scenario:
You are teaching a course like Textural Analysis: Film, Fashion and Material Culture, a course in Harvard’s Department of Visual and Environmental Studies. Students are asked to consider the common language of film and fashion: how they convey identity, create narratives, and shape visual trends. A primary focus is Wong Kar-Wai’s visually arresting film, In the Mood for Love. You ask students to use a video tagging software application (like Motionbox) to tag clips from In the Mood for Love for “fashion elements”: actors’ poses, wardrobes, background settings, colors, makeup, fashion lighting techniques, etc. These tagged elements are then entered into a database and sorted, annotated by the students, compared to cinematic techniques in other films—all toward a new understanding of Wong Kar-Wai’s film and its relationship to fashion.

Another innovation for video analysis, called Metavid, is being hosted by the University of California at Santa Cruz. The project “seeks to capture, stream, archive, and facilitate real-time collective [re]mediation of legislative proceedings” through open source software.

Metavid not only captures the audio and video of US Senate and House floor footage, but also does an OCR capture of the close caption text. All of this metadata is then indexed and made available to the public for searching and downloading in a variety of formats. The project’s Wiki includes a number of compelling Metavid-enabled project ideas and a useful list of similar projects that focus on video annotation, indexing, archiving, “mashups,” and more.

Crowdsourcing Reference

A number of Web search and information providers have been experimenting with providing reference assistance via the Web. These services include Google Answers (discontinued in late December 2006), Yahoo Answers, Windows Live QnA, and the Wikipedia Reference Desk. With the exception of Google’s fee-based model, that hired experienced Researchers to answer questions, most other sites are free and leverage the power of crowdsourcing to cover their virtual reference desks.

Screenshot from the Wikipedia Reference Desk. Click image to visit site.

Some sites, like Yahoo Answers and Windows Live QnA, require users to create individual accounts and have integrated voting tools intended to preserve the quality of the queries and responses. Both of these sites allow users to vote on the “best answer” to a question. Of course, what constitutes the “best answer” may be difficult for users to determine, particularly when several responses to a question might contradict each other.

Virtually all of these reference sites have “codes of conduct” to assist users in asking meaningful questions and answering appropriately. Unfortunately, these guidelines are often ignored by both parties. An exception is Wikipedia’s Reference Desk, where guidelines are posted on the same page used to ask questions. In fact, it is difficult to even ask a question unless you read the guidelines first. Better questions and more detailed answers are clearly the result.

Of all the non-library general reference sites, Wikipedia seems to be the one that is generating the most substantive reference questions. Most of the others seem to attract questions focusing on relationship advice, polling the Internet public on ideas, requesting free technical support for software/computer glitches, etc. The following is a random sampling of questions taken from the Yahoo Answers homepage (the questions are exact quotes):

1. European Cities, if you had to choose 5 citits on the below list which would you visit? The ones the wife and I are thinking about visiting are: London, Edinbraugh, Dublin, Rome, Venice, Paris, Madrid, Barcelona, Amsterdam, Prague, Anthens, Berlin. We are both 26 and newleyweds. We want to take in the sights and party a bit with some relax time, but still on a budget. We will be spending 3-2 nights in each city, so please tell me the five you would visit. Thanks!!

2. I am 23years old. i have completed my B.Sc.(non medical) in2004 in INDIA? ihave a green caard of America now. tell me is it a right time to do computer software enigeneering or i do MBA help me &what i have to do take admission in one of the two.

3. Your theme song? If you were a Superstar or Diva, what would your opening theme song being as you were walking out, (or to represent you) and why? Remember it wouls be something you would hear A LOT. If there isn’t a song for you what would your lyrics for your song be? Dont forget why.

4. How can i have two skype account sigin at the same time?

5. Is Kane the biggest jobber in WWE history? Kane sucks now. They did wrong by taking off his mask. I cant believe he got dismantled in 3 minutes by Khali. Then Regal and Taylor lasted 15 minutes with the guy. MVP is i think 2-1 aganst him. And now he is set to battle mark henry. Does kane see that he is becoming a jobber. He is gonna end up like the Brookly Brawler.

So, can the powerful forces behind Web 2.0, with its focus on user-generated content, be leveraged to provide reliable answers to reference questions as well? Well, there is no doubt that some of these sites are generating great questions and great responses, particularly for technology-related queries. However, a number of challenges exist that keep these sites from being truly reliable sources of information: the use of undisclosed or poor reference sources, contradictory answers to questions, poorly formed questions that are rarely clarified, poorly defined voting systems based on the “best answer,” etc.

Students and researchers looking for accurate information from the most reliable sources will still get much better results more consistently from traditional library reference desks. Skilled librarians who meet face to face with patrons and conduct in-depth reference interviews, who have access to high-quality, peer-reviewed and licensed information resources, and who can teach users how to improve their own search skills, provide a service that I think is essential to research in many fields and the overall quality of scholarship in general.

Still, there are many things that libraries can learn from these reference sites in terms of designing more user-friendly information systems for asking/answering questions, maintaining a reference archive, identifying individuals within a research community willing to share their specialized knowledge, etc. We must also remember that there are many individuals who do not have access to good library reference desks. In these cases, and especially for technology-related topics, something like the Wikipedia Reference Desk is a true blessing.

Improving Image Search

Images are notorious for being among the most challenging objects to search for on the Internet and in library catalogs and databases. Unlike printed works, they typically don’t identify their creator, publication date, or the people, places or events that they visually represent. Creating detailed bibliographic records for images requires painstaking research and expensive cataloging.

 

An image record from Harvard’s Visual Information Access Catalog

A new technology has recently been developed by engineers at UC San Diego that promises to aid the process of searching for images on the Web and in digital projects. Supervised Multiclass Labeling (SML), automatically analyses the content of images, compares it to various “learned” objects and classes, and then assigns searchable labels or keywords to the images. SML can also be used to identify content and generate keywords for different parts of the same image.

In addition to significantly improving image searching on the Web and supplementing the detailed cataloging of image collections by libraries and museums, automated image content analysis technologies like SML could potentially have a number of interesting research applications.

Imagine, for example, being able to perform an image content analysis of Marc Chagall’s paintings, examining the frequency of his religious and folk iconography during different periods of his life. If we think of an artist’s “language” as being expressed through images and symbols, then automated content analysis might some day be used by art historians and researchers as a kind of visual concordance for better understanding art objects, artists, and movements.

The next revolution would then be to apply similar content analysis tools to moving images and videos.

 

Resources:
Article & Video: New Algorithms from UCSD Improve Automated Image Labeling

Article Abstract: Supervised Learning of Semantic Classes for Image Annotation and Retrieval

EdTags: Bookmarking Education

EdTags, created by Adam Seldow, a second year graduate student in the Harvard Graduate School of Education (HGSE), allows researchers to organize, tag, and share their bookmarks, create RSS feeds at the site/user/tag level, view a Facebook-like profile of members within the EdTags community, upload papers, photos, presentations, and more!

Partial screenshot of Edtags homepage. Click image to visit site.

While Edtags is open to the public, 5 HGSE courses, the HGSE library, and 1st year doctoral students received a formal introduction to the tool. In addition to these participants, over 200 more HGSE students as well as participants from various parts of the world share and discover education-related materials on Edtags. The site averages just under 100,000 hits/month, has close to 8,000 pubic and private bookmarks, and about 15,000 tags in its database.

Here are some specific Edtags course/library applications at HGSE:

T-214 Methods of Teaching Mathematics
T-213 Methods of Teaching Social Studies
Both courses conducted “resource fairs” on Edtags using the web-based tool to share, summarize and rate resources for teachers in their respective content areas. Students located and/or authored materials relevant to categories pre-determined by the professor, saved them Edtags and by adding their unique course number as a tag, displayed the recently-shared resources on homepage of the course website via the iSites News RSS reader tool. HGSE graduate students appreciated both the ability to supplement the course syllabus with their favorite resources and the fact that these resources would be available on Edtags post graduation as first year teachers.

T502 Learning Media that Bridge Distance and Time
Professor Chris Dede also used the iSites RSS News Reader tool to display resources his students saved to Edtags bearing the unique course number tag. Students contributed to the body of literature informing the class and further contributed to the field of education through uploading their exemplary course research papers and projects. In addition, students were given the opportunity to work as members of the Edtags project team to gain experience in education research.

T561: Emerging Educational Technologies
Same application as T502 above.

T530: Designing and Producing Media for Education
Students share and discover timely and relevant media tutorials, blog posts, videos and websites. In addition to a few static links on the course website, the students share and access material relevant to the course via the iSites RSS News Reader.

HGSE Gutman Library
Reference Librarians created a group account to share education-related research materials with the HGSE student and faculty community. The librarians supplement the list of static links available on their website with their collection in Edtags. This effort connects researcher with data in an efficient and intuitive manner.

Congratulations to Adam and the Edtags team on building a fantastic tool for educators and researchers!