Re: [CODE4LIB] Looking for two coders to help with discoverability of videos
Hi Kelley - I conducted that usability test on Scherzo and wrote that report so I can answer your questions! I think a work-focused approach can work for users, but we had to scale back on what we assumed users would understand on the search results page. After this test of the system, we changed the search results interface to identify within the works list how many scores and recordings contained that work, so the works list looked more like a facet. The works list then wasn't just a list of titles, but was tied more directly to the recordings/scores result list (which is directly below the works list on the search results page). I do think that some of the testing results we saw reflected how users are used to searching for music in traditional catalogs. While the work is a key concept for musicians, they may have gotten used to the fact that searching for or scanning a results list for a work title often isn't easy (or even possible) in a library catalog so either the title of the album or a person's name is the real key to finding stuff. I think that also might have been part of what threw people off seeing the works listed in the search results. They didn't believe they were seeing titles of songs - they thought they were seeing titles of albums or something that was some sort of physical item. They weren't really sure what it was and so they just skipped that list of things. So adding the info that, for example, a work title is found on 5 recordings/scores really helped to identify the works list as such. Music is kind of unique within FRBR since several works can be involved in a single manifestation (recording or score) and a single work can have many different expressions (different performances by different people of the same work). Other types of resources like books and movies don't often line up with the FRBR model the same way. I can't say for sure whether or not the interface we arrived at after this testing ( http://vfrbr.info/scherzo/) could be used for other work-based resources with a works list serving as a facet to narrow down results, but it seems to be a good use of the FRBR model. Here's an example of a search that I think brings out the strength of what this type of works list can do. Searching in Scherzo for something like symphony no. 5 as Keyword results in several works with that same (or similar) title and lots of recordings and scores that contain expressions of all of the different symphony no. 5 works. The facet nature of showing how many recordings/scores contain that work can help to distinguish which work is the symphony no. 5 you actually want and helps identify that works list as a list of symphony no. 5 works by different composers. I hope this is helpful - it was an interesting project to test these FRBRized search concepts and it would be great to see further experiments with this idea, specifically with non-music resources to see if it can be applied or not. Let me know if you have any more questions about what we did with the Scherzo interface and best of luck on your project! Julie Hardesty Metadata Analyst Metadata Resources Systems Library Technologies Indiana University On Tue, Dec 3, 2013 at 10:58 PM, Kelley McGrath kell...@uoregon.edu wrote: Thanks, Jon. I have seen the Variations work and also talked to Jenn Riley about it. It has definitely influenced me, although we are going in a slightly different direction and moving images have some different needs from music. One thing about Variations that struck me is this paragraph from the usability testing report ( http://www.dlib.indiana.edu/projects/vfrbr/projectDoc/usability/usabilityTest/ScherzoUTestReport.pdf ): There was an assumption among the development team that works would be a window for organizing and narrowing results in a way that users searching for scores and recordings would find useful. One of the main ideas behind FRBR is that the work, or the intellectual entity that is produced by people and is packaged in many forms, is the core information – Scherzo’s interface reflected that organization. 4 (See Appendix E, Fig. 14 for Scherzo’s search results page.) But the participants tended to latch on to a person’s name and search for that name in a particular role. The reasons for this are not completely clear and further discussion follows, but it is worth bearing this finding in mind. Additionally, from the search results page, work results were clicked only 14 times in comparison to items in recordings scores , which were clicked 65 times. Regardless of how the FRBRized data is organized on the back end, the interface needs to reflect the way users want to search, and that might not mean with search results organized by work. Does this mean that a work-focused approach is not actually what users want or need? Does it mean that the work-centered approach needs to be implemented differently in the user interface? Are these results somehow
Re: [CODE4LIB] Looking for two coders to help with discoverability of videos - FRBR
Does this mean that a work-focused approach is not actually what users want or need? Does it mean that the work-centered approach needs to be implemented differently in the user interface? Are these results somehow specific to music? Do they reflect users' familiarity with the typical library catalog and the strategies they've become accustomed to using? FRBR is a wonderful model of our corner of reality. But users aren¹t model-oriented, they are task oriented. They are trying to get stuff done. So the user interface has to make the translation from how systems like to think about the world to how users think about their work. And yes, how users think about their work is shaped by the systems and concepts they¹ve interacted with previously, setting their expectations. But not entirely. To some extent, the Scherzo interface represents an acknowledgement of this after what we learned in the Variations project when trying to make a stepwise FRBRish disambiguation search interface. Here¹s our paper describing that earlier effort: http://www.dlib.indiana.edu/~jenlrile/publications/ecdl2004/ecdl.pdf Mark -- Mark Notess Head, User Experience and Digital Media Services Library Technologies Indiana University Bloomington Libraries +1.812.856.0494 mnot...@iu.edu On 12/6/13, 10:18 AM, Julie Hardesty jlhar...@gmail.com wrote: Hi Kelley - I conducted that usability test on Scherzo and wrote that report so I can answer your questions! I think a work-focused approach can work for users, but we had to scale back on what we assumed users would understand on the search results page. After this test of the system, we changed the search results interface to identify within the works list how many scores and recordings contained that work, so the works list looked more like a facet. The works list then wasn't just a list of titles, but was tied more directly to the recordings/scores result list (which is directly below the works list on the search results page). I do think that some of the testing results we saw reflected how users are used to searching for music in traditional catalogs. While the work is a key concept for musicians, they may have gotten used to the fact that searching for or scanning a results list for a work title often isn't easy (or even possible) in a library catalog so either the title of the album or a person's name is the real key to finding stuff. I think that also might have been part of what threw people off seeing the works listed in the search results. They didn't believe they were seeing titles of songs - they thought they were seeing titles of albums or something that was some sort of physical item. They weren't really sure what it was and so they just skipped that list of things. So adding the info that, for example, a work title is found on 5 recordings/scores really helped to identify the works list as such. Music is kind of unique within FRBR since several works can be involved in a single manifestation (recording or score) and a single work can have many different expressions (different performances by different people of the same work). Other types of resources like books and movies don't often line up with the FRBR model the same way. I can't say for sure whether or not the interface we arrived at after this testing ( http://vfrbr.info/scherzo/) could be used for other work-based resources with a works list serving as a facet to narrow down results, but it seems to be a good use of the FRBR model. Here's an example of a search that I think brings out the strength of what this type of works list can do. Searching in Scherzo for something like symphony no. 5 as Keyword results in several works with that same (or similar) title and lots of recordings and scores that contain expressions of all of the different symphony no. 5 works. The facet nature of showing how many recordings/scores contain that work can help to distinguish which work is the symphony no. 5 you actually want and helps identify that works list as a list of symphony no. 5 works by different composers. I hope this is helpful - it was an interesting project to test these FRBRized search concepts and it would be great to see further experiments with this idea, specifically with non-music resources to see if it can be applied or not. Let me know if you have any more questions about what we did with the Scherzo interface and best of luck on your project! Julie Hardesty Metadata Analyst Metadata Resources Systems Library Technologies Indiana University On Tue, Dec 3, 2013 at 10:58 PM, Kelley McGrath kell...@uoregon.edu wrote: Thanks, Jon. I have seen the Variations work and also talked to Jenn Riley about it. It has definitely influenced me, although we are going in a slightly different direction and moving images have some different needs from music. One thing about Variations that struck me is this paragraph from the usability testing report (
Re: [CODE4LIB] Looking for two coders to help with discoverability of videos
Hi Kelley, If you haven't already, you might want to look at the music score and sound recording FRBRization work done on the Variations-FRBR project here at Indiana University. I'm not sure how directly useful this would be for your work with moving images, but there may be some useful mapping ideas: FRBR XML schemas: http://www.dlib.indiana.edu/projects/vfrbr/schemas/1.1/index.shtml MARC-FRBR mapping specifications: http://www.dlib.indiana.edu/projects/vfrbr/projectDoc/metadata/mappings/spring2010/vfrbrSpring2010mappings.shtml Java FRBRization code and documentation: http://www.dlib.indiana.edu/projects/vfrbr/projectDoc/index.shtml Jon -Original Message- From: Code for Libraries [mailto:CODE4LIB@LISTSERV.ND.EDU] On Behalf Of Kelley McGrath Sent: Tuesday, December 03, 2013 12:35 AM To: CODE4LIB@LISTSERV.ND.EDU Subject: Re: [CODE4LIB] Looking for two coders to help with discoverability of videos Robert, Your work also sounds very interesting and definitely overlaps with some of what we want to do. It seems like a lot of people are trying to get useful format information out of MARC records and it's unfortunate that it is so complicated. I would be very interested to see your logic for determining format and dealing with self-contradictory records. Runtime from the 008 is, as you say, pretty straightforward, but not always filled out and useless if the resource is longer than 999 minutes. It's interesting that you mention identifying directors. We have also been working on a similar, although more generalized, process. We're trying to identify all of the personal and organizational names mentioned in video records and, where possible, their roles. Our existing process is pretty accurate for personal names and for roles in English. It tends to struggle with credits involving multiple corporate bodies and we're working on building a lexicon of non-English terms for common roles. We're also trying to get people to hand-annotate credits to build a corpus to help us improve our process. (Help us out at http://olac-annotator.org/. And if you're willing to be on call to help with translating non-English credits, email me with the language(s) you'd be able to help out with. We also just started a mailing list at https://lists.uoregon.edu/mailman/listinfo/olac-credits) Matching MARC records for moving images with external data sources is also on our radar. Most feature film type material can probably be identified by the attributes you mention: title, original date and director (probably 2 out of 3 would work in most cases). We are also hoping to use these attributes (and possibly others) to cluster records for the same FRBR work. It would be great to talk with you more about this off-list. Kelley kell...@uoregon.edu From: Robert Haschart [rh...@virginia.edu] Sent: Monday, December 02, 2013 10:49 AM To: Code for Libraries Cc: Kelley McGrath Subject: Re: [CODE4LIB] Looking for two coders to help with discoverability of videos Kelley, The work you are proposing is interesting and overlaps somewhat both with work I have already done and with a new project I'm looking into here at UVa. I have been the primary contributor to the Marc4j java project for the past several years and am the creator of the project SolrMarc which extracts data from Marc records based on a customizable specification, to build Solr index records to facilitate rich discovery. Much of my work on creating and improving these projects has been in service of my actual job of creating and maintaining the Solr Index behind our Blacklight-based discovery interface. As a part of that work I have created custom SolrMarc routines that extract the format of items similar to what is described in Example 3, including looking in the leader, 006, 007 and 008 to determine the format as-coded but further looking in the 245 h, 300 and 538 fields to heuristically determine when the format as-coded is incorrect and ought to be overridden. Most of the heuristic determination is targeted towards Video material, and was initiated when I found an item that due to a coding error was listed as a Video in Braille format. Further I have developed a set of custom routines that look more closely at Video items, one of which already extracts the runtime from the 008[18-20] field, To modify it from its current form that currently returns the runtime in minutes, to instead return it as HH:MM as specified in your xls file, and to further handle the edge case of 008[18-20] = 000 to return over 16:39 would literally take about 15 minutes. Another of these custom routines that is more fully-formed, is code for extracting the Director of a video from the Marc record. It examines the contents of the fields 245c, 508a, 500a, 505a, 505t, employing heuristics and targeted natural language processing techniques, to attempt to correctly extract the Director. At
Re: [CODE4LIB] Looking for two coders to help with discoverability of videos
Thanks, Jon. I have seen the Variations work and also talked to Jenn Riley about it. It has definitely influenced me, although we are going in a slightly different direction and moving images have some different needs from music. One thing about Variations that struck me is this paragraph from the usability testing report (http://www.dlib.indiana.edu/projects/vfrbr/projectDoc/usability/usabilityTest/ScherzoUTestReport.pdf): There was an assumption among the development team that works would be a window for organizing and narrowing results in a way that users searching for scores and recordings would find useful. One of the main ideas behind FRBR is that the work, or the intellectual entity that is produced by people and is packaged in many forms, is the core information – Scherzo’s interface reflected that organization. 4 (See Appendix E, Fig. 14 for Scherzo’s search results page.) But the participants tended to latch on to a person’s name and search for that name in a particular role. The reasons for this are not completely clear and further discussion follows, but it is worth bearing this finding in mind. Additionally, from the search results page, work results were clicked only 14 times in comparison to items in recordings scores , which were clicked 65 times. Regardless of how the FRBRized data is organized on the back end, the interface needs to reflect the way users want to search, and that might not mean with search results organized by work. Does this mean that a work-focused approach is not actually what users want or need? Does it mean that the work-centered approach needs to be implemented differently in the user interface? Are these results somehow specific to music? Do they reflect users' familiarity with the typical library catalog and the strategies they've become accustomed to using? It does suggest to me that there should be more studies on how users interact with FRBRized data (and not just the clustering that so many discovery interfaces do now, but real FRBR-based data) and how FRBRized data is best presented. Kelley On Tue, Dec 3, 2013 at 11:35 AM, Dunn, Jon William Butcher j...@iu.edumailto:j...@iu.edu wrote: Hi Kelley, If you haven't already, you might want to look at the music score and sound recording FRBRization work done on the Variations-FRBR project here at Indiana University. I'm not sure how directly useful this would be for your work with moving images, but there may be some useful mapping ideas: FRBR XML schemas: http://www.dlib.indiana.edu/projects/vfrbr/schemas/1.1/index.shtml MARC-FRBR mapping specifications: http://www.dlib.indiana.edu/projects/vfrbr/projectDoc/metadata/mappings/spring2010/vfrbrSpring2010mappings.shtml Java FRBRization code and documentation: http://www.dlib.indiana.edu/projects/vfrbr/projectDoc/index.shtml Jon
Re: [CODE4LIB] Looking for two coders to help with discoverability of videos
Is it out of the question to extract technical metadata from the audiovisual materials themselves (via MediaInfo et al)? It would minimize the amount of MARC that needs to be processed and give more accurate/complete data than relying on old cataloging records. On Mon, Dec 2, 2013 at 12:37 AM, Kelley McGrath kell...@uoregon.edu wrote: I wanted to follow up on my previous post with a couple points. 1. This is probably too late for anybody thinking about applying, but I thought there may be some general interest. I have put up some more detailed specifications about what I am hoping to do at http://pages.uoregon.edu/kelleym/miw/. Data extraction overview.doc is the general overview and the other files contain supporting documents. 2. I replied some time ago to Heather's offer below about her website that will connect researchers with volunteer software developers. I have to admit that looking for volunteer software developers had not really occurred to me. However, I do have additional things that I would like to do for which I currently have no funding so if you would be interested in volunteering in the future, let me know. Kelley kell...@uoregon.edu On Tue, Nov 12, 2013 at 6:33 PM, Heather Claxton claxt...@gmail.com mailto:claxt...@gmail.com wrote: Hi Kelley, I might be able to help in your search. I'm in the process of starting a website that connects academic researchers with volunteer software developers. I'm looking for people to post programming projects on the website once it's launched in late January. I realize that may be a little late for you, but perhaps the project you mentioned in your PS (clustering based on title, name, date ect.) would be perfect? The one caveat is that the website is targeting software developers who wish to volunteer. Anyway, if you're interested in posting, please send me an e-mail at sciencesolved2...@gmail.commailto:sciencesolved2...@gmail.com I would greatly appreciate it. Oh and of course it would be free to post :) Best of luck in your hiring process, Heather Claxton-Douglas On Mon, Nov 11, 2013 at 9:58 PM, Kelley McGrath kell...@uoregon.edu mailto:kell...@uoregon.edu wrote: I have a small amount of money to work with and am looking for two people to help with extracting data from MARC records as described below. This is part of a larger project to develop a FRBR-based data store and discovery interface for moving images. Our previous work includes a consideration of the feasibility of the project from a cataloging perspective ( http://www.olacinc.org/drupal/?q=node/27), a prototype end-user interface (https://blazing-sunset-24.heroku.com/, https://blazing-sunset-24.heroku.com/page/about) and a web form to crowdsource the parsing of movie credits ( http://olac-annotator.org/#/about). Planned work period: six months beginning around the second week of December (I can be somewhat flexible on the dates if you want to wait and start after the New Year) Payment: flat sum of $2500 upon completion of the work Required skills and knowledge: * Familiarity with the MARC 21 bibliographic format * Familiarity with Natural Language Processing concepts (or willingness to learn) * Experience with Java, Python, and/or Ruby programming languages Description of work: Use language and text processing tools and provided strategies to write code to extract and normalize data in existing MARC bibliographic records for moving images. Refine code based on feedback from analysis of results obtained with a sample dataset. Data to be extracted: Tasks for Position 1: Titles (including the main title of the video, uniform titles, variant titles, series titles, television program titles and titles of contents) Authors and titles of related works on which an adaptation is based Duration Color Sound vs. silent Tasks for Position 2: Format (DVD, VHS, film, online, etc.) Original language Country of production Aspect ratio Flag for whether a record represents multiple works or not We have already done some work with dates, names and roles and have a framework to work in. I have the basic logic for the data extraction processes, but expect to need some iteration to refine these strategies. To apply please send me an email at kelleym@uoregon explaining why you are interested in this project, what relevant experience you would bring and any other reasons why I should hire you. If you have a preference for position 1 or 2, let me know (it's not necessary to have a preference). The deadline for applications is Monday, December 2, 2013. Let me know if you have any questions. Thank you for your consideration. Kelley PS In the near future, I will also be looking for someone to help with work clustering based on title, name, date and identifier data from MARC records. This will not involve any direct
Re: [CODE4LIB] Looking for two coders to help with discoverability of videos
Is it out of the question to extract technical metadata from the audiovisual materials themselves (via MediaInfo et al)? One of the things that absolutely blows my mind is the widespread practice of hand typing this stuff into records. Aside from an obvious opportunity to introduce errors/inconsistencies, many libraries record details for the archival versions rather than the access versions actually provided. So patrons see a description for what they're not getting... Just for the heck of it, sometime last year I scanned thousands of objects and their descriptions to see how close they were. Like an idiot, I didn't write up what I learned because I was just trying to satisfy my own curiosity. However, the takeaway I got from the exercise was that the embedded info is so much better than the hand keyed stuff that you'd be nuts to consider the latter as authoritative. Curiously, I did find cases where the embedded info was clearly incorrect. I can only guess that was manually edited. kyle
Re: [CODE4LIB] Looking for two coders to help with discoverability of videos
I would have to agree with this where the data exists. The data captured by digital cameras these days can be incredibly extensive and thorough. Given this, I recently started exposing this data for all of the 8,000 photos I now have on my photos web site http://FreeLargePhotos.com/ . There is now a link on the page for an individual photo that a user can click on that will pull out the data dynamically from the image file and display it in plain text. Here is a random example: http://freelargephotos.com/photos/003171/exif.txt The tricky bit is of course where the photo is actually scanned from a slide, which of course plays havoc with items such as the creation date. So depending on the exact situation your mileage may vary, but the basic principle stands -- if you can allow a machine to capture the metadata then by all means let it. Roy On Mon, Dec 2, 2013 at 9:06 AM, Kyle Banerjee kyle.baner...@gmail.comwrote: Is it out of the question to extract technical metadata from the audiovisual materials themselves (via MediaInfo et al)? One of the things that absolutely blows my mind is the widespread practice of hand typing this stuff into records. Aside from an obvious opportunity to introduce errors/inconsistencies, many libraries record details for the archival versions rather than the access versions actually provided. So patrons see a description for what they're not getting... Just for the heck of it, sometime last year I scanned thousands of objects and their descriptions to see how close they were. Like an idiot, I didn't write up what I learned because I was just trying to satisfy my own curiosity. However, the takeaway I got from the exercise was that the embedded info is so much better than the hand keyed stuff that you'd be nuts to consider the latter as authoritative. Curiously, I did find cases where the embedded info was clearly incorrect. I can only guess that was manually edited. kyle
Re: [CODE4LIB] Looking for two coders to help with discoverability of videos - Embedded Metadata
I've been working with embedded metadata for some years and there are great tools out there for embedding, extracting and reusing metadata (technical, administrative, and descriptive). The tools allow for batch data entry, use metadata schema or standards. As a digital archivist whose job is to take in lots of this digitized content that generally has no context or that context is lost or misplaced, I wholly advocate for embedding metadata. There are consumer products that can then expose this metadata so that it doesn't have to be retyped again and again. What gets my goat is when I hear folks belabor the effort but don't talk about the rewards and opportunities that embedding metadata can bring. Forthcoming use cases from The Royal Library in Denmark about mass digitization and embedding metadata as well as using the Exif / IPTC Extension for describing the content in image files. There's also work being done with video and audio and CAD files. Check out these resources on Embedded Metadata from the VRA Embedded Metadata Working Group (Greg Reser, Chair): About Embedded Metadata: http://metadatadeluxe.pbworks.com/w/page/62407805/Concepts http://metadatadeluxe.pbworks.com/w/page/20792256/Other%20Organizations Case Studies: http://metadatadeluxe.pbworks.com/w/page/62407826/Communities Okay, I'll step off my soap box now... Kari -Original Message- From: Code for Libraries [mailto:CODE4LIB@listserv.nd.edu] On Behalf Of Kyle Banerjee Sent: Monday, December 02, 2013 12:06 PM To: CODE4LIB@listserv.nd.edu Subject: Re: [CODE4LIB] Looking for two coders to help with discoverability of videos Is it out of the question to extract technical metadata from the audiovisual materials themselves (via MediaInfo et al)? One of the things that absolutely blows my mind is the widespread practice of hand typing this stuff into records. Aside from an obvious opportunity to introduce errors/inconsistencies, many libraries record details for the archival versions rather than the access versions actually provided. So patrons see a description for what they're not getting... Just for the heck of it, sometime last year I scanned thousands of objects and their descriptions to see how close they were. Like an idiot, I didn't write up what I learned because I was just trying to satisfy my own curiosity. However, the takeaway I got from the exercise was that the embedded info is so much better than the hand keyed stuff that you'd be nuts to consider the latter as authoritative. Curiously, I did find cases where the embedded info was clearly incorrect. I can only guess that was manually edited. kyle
Re: [CODE4LIB] Looking for two coders to help with discoverability of videos
Kelley, The work you are proposing is interesting and overlaps somewhat both with work I have already done and with a new project I'm looking into here at UVa. I have been the primary contributor to the Marc4j java project for the past several years and am the creator of the project SolrMarc which extracts data from Marc records based on a customizable specification, to build Solr index records to facilitate rich discovery. Much of my work on creating and improving these projects has been in service of my actual job of creating and maintaining the Solr Index behind our Blacklight-based discovery interface. As a part of that work I have created custom SolrMarc routines that extract the format of items similar to what is described in Example 3, including looking in the leader, 006, 007 and 008 to determine the format as-coded but further looking in the 245 h, 300 and 538 fields to heuristically determine when the format as-coded is incorrect and ought to be overridden. Most of the heuristic determination is targeted towards Video material, and was initiated when I found an item that due to a coding error was listed as a Video in Braille format. Further I have developed a set of custom routines that look more closely at Video items, one of which already extracts the runtime from the 008[18-20] field, To modify it from its current form that currently returns the runtime in minutes, to instead return it as HH:MM as specified in your xls file, and to further handle the edge case of 008[18-20] = 000 to return over 16:39 would literally take about 15 minutes. Another of these custom routines that is more fully-formed, is code for extracting the Director of a video from the Marc record. It examines the contents of the fields 245c, 508a, 500a, 505a, 505t, employing heuristics and targeted natural language processing techniques, to attempt to correctly extract the Director. At this point I believe it achieves better results than a careful cataloger would achieve, even one who specializes in film and video. The other project I have just started investigating is an effort to create and/or flesh out Marc records for video items based on heuristic matching of title and director and date with data returned from publicly-accessible movie information sites. This more recent work may not be relevant to your needs but the custom extraction routines seem directly applicable to your goals, and may also provide a template that may make your other goals more easily achievable. -Robert Haschart On 12/2/2013 12:37 AM, Kelley McGrath wrote: I wanted to follow up on my previous post with a couple points. 1. This is probably too late for anybody thinking about applying, but I thought there may be some general interest. I have put up some more detailed specifications about what I am hoping to do at http://pages.uoregon.edu/kelleym/miw/. Data extraction overview.doc is the general overview and the other files contain supporting documents. 2. I replied some time ago to Heather's offer below about her website that will connect researchers with volunteer software developers. I have to admit that looking for volunteer software developers had not really occurred to me. However, I do have additional things that I would like to do for which I currently have no funding so if you would be interested in volunteering in the future, let me know. Kelley kell...@uoregon.edu On Tue, Nov 12, 2013 at 6:33 PM, Heather Claxtonclaxt...@gmail.commailto:claxt...@gmail.com wrote: Hi Kelley, I might be able to help in your search. I'm in the process of starting a website that connects academic researchers with volunteer software developers. I'm looking for people to post programming projects on the website once it's launched in late January. I realize that may be a little late for you, but perhaps the project you mentioned in your PS (clustering based on title, name, date ect.) would be perfect? The one caveat is that the website is targeting software developers who wish to volunteer. Anyway, if you're interested in posting, please send me an e-mail at sciencesolved2...@gmail.commailto:sciencesolved2...@gmail.com I would greatly appreciate it. Oh and of course it would be free to post :) Best of luck in your hiring process, Heather Claxton-Douglas On Mon, Nov 11, 2013 at 9:58 PM, Kelley McGrathkell...@uoregon.edumailto:kell...@uoregon.edu wrote: I have a small amount of money to work with and am looking for two people to help with extracting data from MARC records as described below. This is part of a larger project to develop a FRBR-based data store and discovery interface for moving images. Our previous work includes a consideration of the feasibility of the project from a cataloging perspective ( http://www.olacinc.org/drupal/?q=node/27), a prototype end-user interface (https://blazing-sunset-24.heroku.com/,
Re: [CODE4LIB] Looking for two coders to help with discoverability of videos
Well, that would be much easier, but most of what I am working with are records for physical items (DVD, VHS, film) or licensed streaming video. The sample records are also not all UO records so I don't necessarily even have access to the source material (our goal is to build a general purpose tool). So I think I am stuck with extracting from MARC. We should be able to get data for some resources by matching the MARC up with external data sources. That won't work for everything, though, so we want to make the process of extracting data from MARC as effective as possible. Kelley On Mon, Dec 2, 2013 at 7:03 AM, Alexander Duryee alexanderdur...@gmail.commailto:alexanderdur...@gmail.com wrote: Is it out of the question to extract technical metadata from the audiovisual materials themselves (via MediaInfo et al)? It would minimize the amount of MARC that needs to be processed and give more accurate/complete data than relying on old cataloging records.
Re: [CODE4LIB] Looking for two coders to help with discoverability of videos
Robert, Your work also sounds very interesting and definitely overlaps with some of what we want to do. It seems like a lot of people are trying to get useful format information out of MARC records and it's unfortunate that it is so complicated. I would be very interested to see your logic for determining format and dealing with self-contradictory records. Runtime from the 008 is, as you say, pretty straightforward, but not always filled out and useless if the resource is longer than 999 minutes. It's interesting that you mention identifying directors. We have also been working on a similar, although more generalized, process. We're trying to identify all of the personal and organizational names mentioned in video records and, where possible, their roles. Our existing process is pretty accurate for personal names and for roles in English. It tends to struggle with credits involving multiple corporate bodies and we're working on building a lexicon of non-English terms for common roles. We're also trying to get people to hand-annotate credits to build a corpus to help us improve our process. (Help us out at http://olac-annotator.org/. And if you're willing to be on call to help with translating non-English credits, email me with the language(s) you'd be able to help out with. We also just started a mailing list at https://lists.uoregon.edu/mailman/listinfo/olac-credits) Matching MARC records for moving images with external data sources is also on our radar. Most feature film type material can probably be identified by the attributes you mention: title, original date and director (probably 2 out of 3 would work in most cases). We are also hoping to use these attributes (and possibly others) to cluster records for the same FRBR work. It would be great to talk with you more about this off-list. Kelley kell...@uoregon.edu From: Robert Haschart [rh...@virginia.edu] Sent: Monday, December 02, 2013 10:49 AM To: Code for Libraries Cc: Kelley McGrath Subject: Re: [CODE4LIB] Looking for two coders to help with discoverability of videos Kelley, The work you are proposing is interesting and overlaps somewhat both with work I have already done and with a new project I'm looking into here at UVa. I have been the primary contributor to the Marc4j java project for the past several years and am the creator of the project SolrMarc which extracts data from Marc records based on a customizable specification, to build Solr index records to facilitate rich discovery. Much of my work on creating and improving these projects has been in service of my actual job of creating and maintaining the Solr Index behind our Blacklight-based discovery interface. As a part of that work I have created custom SolrMarc routines that extract the format of items similar to what is described in Example 3, including looking in the leader, 006, 007 and 008 to determine the format as-coded but further looking in the 245 h, 300 and 538 fields to heuristically determine when the format as-coded is incorrect and ought to be overridden. Most of the heuristic determination is targeted towards Video material, and was initiated when I found an item that due to a coding error was listed as a Video in Braille format. Further I have developed a set of custom routines that look more closely at Video items, one of which already extracts the runtime from the 008[18-20] field, To modify it from its current form that currently returns the runtime in minutes, to instead return it as HH:MM as specified in your xls file, and to further handle the edge case of 008[18-20] = 000 to return over 16:39 would literally take about 15 minutes. Another of these custom routines that is more fully-formed, is code for extracting the Director of a video from the Marc record. It examines the contents of the fields 245c, 508a, 500a, 505a, 505t, employing heuristics and targeted natural language processing techniques, to attempt to correctly extract the Director. At this point I believe it achieves better results than a careful cataloger would achieve, even one who specializes in film and video. The other project I have just started investigating is an effort to create and/or flesh out Marc records for video items based on heuristic matching of title and director and date with data returned from publicly-accessible movie information sites. This more recent work may not be relevant to your needs but the custom extraction routines seem directly applicable to your goals, and may also provide a template that may make your other goals more easily achievable. -Robert Haschart
Re: [CODE4LIB] Looking for two coders to help with discoverability of videos
I wanted to follow up on my previous post with a couple points. 1. This is probably too late for anybody thinking about applying, but I thought there may be some general interest. I have put up some more detailed specifications about what I am hoping to do at http://pages.uoregon.edu/kelleym/miw/. Data extraction overview.doc is the general overview and the other files contain supporting documents. 2. I replied some time ago to Heather's offer below about her website that will connect researchers with volunteer software developers. I have to admit that looking for volunteer software developers had not really occurred to me. However, I do have additional things that I would like to do for which I currently have no funding so if you would be interested in volunteering in the future, let me know. Kelley kell...@uoregon.edu On Tue, Nov 12, 2013 at 6:33 PM, Heather Claxton claxt...@gmail.commailto:claxt...@gmail.com wrote: Hi Kelley, I might be able to help in your search. I'm in the process of starting a website that connects academic researchers with volunteer software developers. I'm looking for people to post programming projects on the website once it's launched in late January. I realize that may be a little late for you, but perhaps the project you mentioned in your PS (clustering based on title, name, date ect.) would be perfect? The one caveat is that the website is targeting software developers who wish to volunteer. Anyway, if you're interested in posting, please send me an e-mail at sciencesolved2...@gmail.commailto:sciencesolved2...@gmail.com I would greatly appreciate it. Oh and of course it would be free to post :) Best of luck in your hiring process, Heather Claxton-Douglas On Mon, Nov 11, 2013 at 9:58 PM, Kelley McGrath kell...@uoregon.edumailto:kell...@uoregon.edu wrote: I have a small amount of money to work with and am looking for two people to help with extracting data from MARC records as described below. This is part of a larger project to develop a FRBR-based data store and discovery interface for moving images. Our previous work includes a consideration of the feasibility of the project from a cataloging perspective ( http://www.olacinc.org/drupal/?q=node/27), a prototype end-user interface (https://blazing-sunset-24.heroku.com/, https://blazing-sunset-24.heroku.com/page/about) and a web form to crowdsource the parsing of movie credits ( http://olac-annotator.org/#/about). Planned work period: six months beginning around the second week of December (I can be somewhat flexible on the dates if you want to wait and start after the New Year) Payment: flat sum of $2500 upon completion of the work Required skills and knowledge: * Familiarity with the MARC 21 bibliographic format * Familiarity with Natural Language Processing concepts (or willingness to learn) * Experience with Java, Python, and/or Ruby programming languages Description of work: Use language and text processing tools and provided strategies to write code to extract and normalize data in existing MARC bibliographic records for moving images. Refine code based on feedback from analysis of results obtained with a sample dataset. Data to be extracted: Tasks for Position 1: Titles (including the main title of the video, uniform titles, variant titles, series titles, television program titles and titles of contents) Authors and titles of related works on which an adaptation is based Duration Color Sound vs. silent Tasks for Position 2: Format (DVD, VHS, film, online, etc.) Original language Country of production Aspect ratio Flag for whether a record represents multiple works or not We have already done some work with dates, names and roles and have a framework to work in. I have the basic logic for the data extraction processes, but expect to need some iteration to refine these strategies. To apply please send me an email at kelleym@uoregon explaining why you are interested in this project, what relevant experience you would bring and any other reasons why I should hire you. If you have a preference for position 1 or 2, let me know (it's not necessary to have a preference). The deadline for applications is Monday, December 2, 2013. Let me know if you have any questions. Thank you for your consideration. Kelley PS In the near future, I will also be looking for someone to help with work clustering based on title, name, date and identifier data from MARC records. This will not involve any direct interaction with MARC. Kelley McGrath Metadata Management Librarian University of Oregon Libraries 541-346-8232tel:541-346-8232 kell...@uoregon.edumailto:kell...@uoregon.edu
Re: [CODE4LIB] Looking for two coders to help with discoverability of videos
Hi Kelley, Thanks for posting this. When I began work on jobs.code4lib.org I was hoping it would encourage people to post short term contracts. The thought being that it may be easier for some institutions to find money for projects than full-time staff, and it could encourage more open source collaboration between organizations, similar to what the Hydra Project are doing. So, I added your post to jobs.code4lib.org [1]. Ordinarily the person who publishes a job posting is the only one who can edit it. But if you would like to make any changes to it please let me know and I’ll make you the editor. Incidentally I was curious about your decision to hire two programmers to do what appears to be a very similar task. Was your intent to have two implementations to compare to see which you liked better? Were the two developers supposed to work together or separately? //Ed [1] http://jobs.code4lib.org/job/10658/ On Nov 11, 2013, at 10:58 PM, Kelley McGrath kell...@uoregon.edu wrote: I have a small amount of money to work with and am looking for two people to help with extracting data from MARC records as described below. This is part of a larger project to develop a FRBR-based data store and discovery interface for moving images. Our previous work includes a consideration of the feasibility of the project from a cataloging perspective (http://www.olacinc.org/drupal/?q=node/27), a prototype end-user interface (https://blazing-sunset-24.heroku.com/, https://blazing-sunset-24.heroku.com/page/about) and a web form to crowdsource the parsing of movie credits (http://olac-annotator.org/#/about). Planned work period: six months beginning around the second week of December (I can be somewhat flexible on the dates if you want to wait and start after the New Year) Payment: flat sum of $2500 upon completion of the work Required skills and knowledge: * Familiarity with the MARC 21 bibliographic format * Familiarity with Natural Language Processing concepts (or willingness to learn) * Experience with Java, Python, and/or Ruby programming languages Description of work: Use language and text processing tools and provided strategies to write code to extract and normalize data in existing MARC bibliographic records for moving images. Refine code based on feedback from analysis of results obtained with a sample dataset. Data to be extracted: Tasks for Position 1: Titles (including the main title of the video, uniform titles, variant titles, series titles, television program titles and titles of contents) Authors and titles of related works on which an adaptation is based Duration Color Sound vs. silent Tasks for Position 2: Format (DVD, VHS, film, online, etc.) Original language Country of production Aspect ratio Flag for whether a record represents multiple works or not We have already done some work with dates, names and roles and have a framework to work in. I have the basic logic for the data extraction processes, but expect to need some iteration to refine these strategies. To apply please send me an email at kelleym@uoregon explaining why you are interested in this project, what relevant experience you would bring and any other reasons why I should hire you. If you have a preference for position 1 or 2, let me know (it's not necessary to have a preference). The deadline for applications is Monday, December 2, 2013. Let me know if you have any questions. Thank you for your consideration. Kelley PS In the near future, I will also be looking for someone to help with work clustering based on title, name, date and identifier data from MARC records. This will not involve any direct interaction with MARC. Kelley McGrath Metadata Management Librarian University of Oregon Libraries 541-346-8232 kell...@uoregon.edu
Re: [CODE4LIB] Looking for two coders to help with discoverability of videos
+1 for what I know of Avalon Media service -- Al Matthews Software Developer, Digital Services Unit Atlanta University Center, Robert W. Woodruff Library email: amatth...@auctr.edu; office: 1 404 978 2057 On 11/12/13 8:21 AM, Edward Summers e...@pobox.com wrote: Hi Kelley, Thanks for posting this. When I began work on jobs.code4lib.org I was hoping it would encourage people to post short term contracts. The thought being that it may be easier for some institutions to find money for projects than full-time staff, and it could encourage more open source collaboration between organizations, similar to what the Hydra Project are doing. So, I added your post to jobs.code4lib.org [1]. Ordinarily the person who publishes a job posting is the only one who can edit it. But if you would like to make any changes to it please let me know and I’ll make you the editor. Incidentally I was curious about your decision to hire two programmers to do what appears to be a very similar task. Was your intent to have two implementations to compare to see which you liked better? Were the two developers supposed to work together or separately? //Ed [1] http://jobs.code4lib.org/job/10658/ On Nov 11, 2013, at 10:58 PM, Kelley McGrath kell...@uoregon.edu wrote: I have a small amount of money to work with and am looking for two people to help with extracting data from MARC records as described below. This is part of a larger project to develop a FRBR-based data store and discovery interface for moving images. Our previous work includes a consideration of the feasibility of the project from a cataloging perspective (http://www.olacinc.org/drupal/?q=node/27), a prototype end-user interface (https://blazing-sunset-24.heroku.com/, https://blazing-sunset-24.heroku.com/page/about) and a web form to crowdsource the parsing of movie credits (http://olac-annotator.org/#/about). Planned work period: six months beginning around the second week of December (I can be somewhat flexible on the dates if you want to wait and start after the New Year) Payment: flat sum of $2500 upon completion of the work Required skills and knowledge: * Familiarity with the MARC 21 bibliographic format * Familiarity with Natural Language Processing concepts (or willingness to learn) * Experience with Java, Python, and/or Ruby programming languages Description of work: Use language and text processing tools and provided strategies to write code to extract and normalize data in existing MARC bibliographic records for moving images. Refine code based on feedback from analysis of results obtained with a sample dataset. Data to be extracted: Tasks for Position 1: Titles (including the main title of the video, uniform titles, variant titles, series titles, television program titles and titles of contents) Authors and titles of related works on which an adaptation is based Duration Color Sound vs. silent Tasks for Position 2: Format (DVD, VHS, film, online, etc.) Original language Country of production Aspect ratio Flag for whether a record represents multiple works or not We have already done some work with dates, names and roles and have a framework to work in. I have the basic logic for the data extraction processes, but expect to need some iteration to refine these strategies. To apply please send me an email at kelleym@uoregon explaining why you are interested in this project, what relevant experience you would bring and any other reasons why I should hire you. If you have a preference for position 1 or 2, let me know (it's not necessary to have a preference). The deadline for applications is Monday, December 2, 2013. Let me know if you have any questions. Thank you for your consideration. Kelley PS In the near future, I will also be looking for someone to help with work clustering based on title, name, date and identifier data from MARC records. This will not involve any direct interaction with MARC. Kelley McGrath Metadata Management Librarian University of Oregon Libraries 541-346-8232 kell...@uoregon.edu
Re: [CODE4LIB] Looking for two coders to help with discoverability of videos
Hi Kelley, I might be able to help in your search. I'm in the process of starting a website that connects academic researchers with volunteer software developers. I'm looking for people to post programming projects on the website once it's launched in late January. I realize that may be a little late for you, but perhaps the project you mentioned in your PS (clustering based on title, name, date ect.) would be perfect? The one caveat is that the website is targeting software developers who wish to volunteer. Anyway, if you're interested in posting, please send me an e-mail at sciencesolved2...@gmail.comI would greatly appreciate it. Oh and of course it would be free to post :) Best of luck in your hiring process, Heather Claxton-Douglas On Mon, Nov 11, 2013 at 9:58 PM, Kelley McGrath kell...@uoregon.edu wrote: I have a small amount of money to work with and am looking for two people to help with extracting data from MARC records as described below. This is part of a larger project to develop a FRBR-based data store and discovery interface for moving images. Our previous work includes a consideration of the feasibility of the project from a cataloging perspective ( http://www.olacinc.org/drupal/?q=node/27), a prototype end-user interface (https://blazing-sunset-24.heroku.com/, https://blazing-sunset-24.heroku.com/page/about) and a web form to crowdsource the parsing of movie credits ( http://olac-annotator.org/#/about). Planned work period: six months beginning around the second week of December (I can be somewhat flexible on the dates if you want to wait and start after the New Year) Payment: flat sum of $2500 upon completion of the work Required skills and knowledge: * Familiarity with the MARC 21 bibliographic format * Familiarity with Natural Language Processing concepts (or willingness to learn) * Experience with Java, Python, and/or Ruby programming languages Description of work: Use language and text processing tools and provided strategies to write code to extract and normalize data in existing MARC bibliographic records for moving images. Refine code based on feedback from analysis of results obtained with a sample dataset. Data to be extracted: Tasks for Position 1: Titles (including the main title of the video, uniform titles, variant titles, series titles, television program titles and titles of contents) Authors and titles of related works on which an adaptation is based Duration Color Sound vs. silent Tasks for Position 2: Format (DVD, VHS, film, online, etc.) Original language Country of production Aspect ratio Flag for whether a record represents multiple works or not We have already done some work with dates, names and roles and have a framework to work in. I have the basic logic for the data extraction processes, but expect to need some iteration to refine these strategies. To apply please send me an email at kelleym@uoregon explaining why you are interested in this project, what relevant experience you would bring and any other reasons why I should hire you. If you have a preference for position 1 or 2, let me know (it's not necessary to have a preference). The deadline for applications is Monday, December 2, 2013. Let me know if you have any questions. Thank you for your consideration. Kelley PS In the near future, I will also be looking for someone to help with work clustering based on title, name, date and identifier data from MARC records. This will not involve any direct interaction with MARC. Kelley McGrath Metadata Management Librarian University of Oregon Libraries 541-346-8232 kell...@uoregon.edu