hybl@mbph.UUCP (Albert Hybl Dept of Biophysics SM) (12/19/89)
Dear Santa, The Internet has been providing useful services to the academic community. It provides a vehicle for exchange of electronic mail; the usenet is an international bulletin board for communication with a wide audience; it disseminates and archives public domain software; bionet.journal.contents is the fledgling equivalent of an electronic current contents; the bionet.sci-resources announces funding opportunities and changes in NIH regulations. Contributors to the bionet.molbio.genbank news group are advocating that the Internet be used for the electronic transfer of current and archived gene sequences. I agree! However, the Internet could also be used for the dissemination of protein sequences, crystallographic data from small molecules and macromolecules, bibliographic abstracts such as Med-Line or Chem Abstracts and a host of other information resources. There appears to be a need to establish electronic research libraries. Slip into any bricked academic library and you can find shelf after shelf devoted to various publications such as Chem Abstracts and Index Medicus, a few terminals used for accessing Med-Line, and several PCs each dedicated to a single minded task. Faculty and students use these facilities to keep current on certain subjects or to locate reference to information for a research project. Surely it would be much more efficient to receive electronic mail or to login and read from the current postings to info.groups based the user's interest profile. The electronic research libraries could use the gigabyte or multiple terabyte optical storage devices; these devices are considerably less expensive than building more brick libraries. What are the advantages? o Rapid posting of new entries The public domain software already developed for the usenet can be used to transfer compressed batches of information between the libraries. The generators of the information can post directly to the network eliminating publication delays or tape/disk production and distribution problems. o User profile subject selection (.medrc or .chemrc) The National Library of Medicine publishes a tree structure for its Medical Subject Headings; hence, just as a user defines the news group he wishes to read in a .newsrc file, he could define his interest profile in a .medrc or .chemrc file. Again the public domain software of the usenet could be used. o Archiving the entries Archiving is essential. Once it becomes obvious that the PCs located in brick libraries can not provide the breath of service that is technically possible, then perhaps enlightened administrators will begin to equip and staff the facilities needed to accomplish the job. Software like the M. Lesk Refer programs can be used to maintain links to the archived information. o User retrieval of archived entries An UNIX shell program like lookbib or seekbib can be used to extract relevant reference from an M. Lesk archive. Because it is unlikely that all users will be able to ftp directly to the computer system that contains the archive, they should be able to issue a control command that would cause the remote execution of a seekbib shell with the results of the search returned via e-mail. Although there is plenty of room for improvements, software for all the proposed services exist. Technically both hardware and software exists to create electronic research libraries. So, Santa, please give our administrators the wisdom to speedily provide the services. Thank you, ---------------------------------------------------------------------- Albert Hybl, PhD. UUCP Office: uunet!mimsy!mbph!hybl Department of Biophysics Home: uunet!mimsy!mbph!hybl!ah University of Maryland Bitnet: hybl@umbc1 CoSy: ahybl School of Medicine Phone Office: (301) 328-7940 Baltimore, MD 21201 Home: (301) 243-1710 ---------------------------------------------------------------------- Responders--DO NOT USE: hybl@cs.umd.edu or ah@cs.umd.edu
kristoff@genbank.BIO.NET (David Kristofferson) (12/20/89)
Albert, You may be relieved to note that many people at the NIH read the bionet.* newsgroups, so if Santa lives at the NIH then he/she may be listening. The usual problem, though, is who is going to pay for all of these wonderful things. Perhaps an answer might be found in the long term savings on library shelf space, buildings, etc. This is more than just an NIH jurisdictional issue unfortunately. It may be of interest to note that Senator Gore is pushing a national networking initiative and this may be a good place to begin rallying support. Ho, ho, ho! -- Sincerely, Dave Kristofferson GenBank On-line Service Manager kristoff@genbank.bio.net
sachs@tanner.berkeley.edu (Rainer Sachs) (12/20/89)
In article <617@mbph.UUCP> hybl@mbph.UUCP (Albert Hybl Dept of Biophysics SM) writes: > >Dear Santa,... >There appears to be a need to establish electronic research libraries. >... >An UNIX shell program like lookbib or seekbib can be used to extract >relevant reference from an M. Lesk archive. Because it is unlikely >that all users will be able to ftp directly to the computer system >that contains the archive, they should be able to issue a control >command that would cause the remote execution of a seekbib shell >with the results of the search returned via e-mail. ... Is this advocating non-interactive searches? Speaking for myself, I think a non-interactive search would be virtually useless. I have lot of experience searching medline. I can't recall *ever* getting the search exactly right the first time. When I do get it right, I always need to pick and choose among abstracts in order to get the ones I want without hopelessly cluttering up my own reference files, duplicating, etc. Assuming Santa has limited funds, I would urge him to put them only into programs and situations where interactive searches are possible. How does the net feel?
karish@forel.stanford.edu (Chuck Karish) (12/20/89)
[ Note the `Followup-To:' header. This is an exciting topic as far as data sharing goes, but it's probably a tangential thread for most of the newsgroups in the original distribution. --crk ] In article <1989Dec20.011113.724@agate.berkeley.edu> sachs@tanner.UUCP (Rainer Sachs) wrote: >In article <617@mbph.UUCP> hybl@mbph.UUCP (Albert Hybl Dept of Biophysics SM) writes: >>Dear Santa,... >>There appears to be a need to establish electronic research libraries. >>... >>An UNIX shell program like lookbib or seekbib can be used to extract >>relevant reference from an M. Lesk archive. Because it is unlikely >>that all users will be able to ftp directly to the computer system >>that contains the archive, they should be able to issue a control >>command that would cause the remote execution of a seekbib shell >>with the results of the search returned via e-mail. ... > >Is this advocating non-interactive searches? So it would seem. >Speaking for myself, I think a non-interactive search would be >virtually useless. I have lot of experience searching medline. >I can't recall *ever* getting the search exactly right the first >time. When I do get it right, I always need to pick and >choose among abstracts in order to get the ones I want without >hopelessly cluttering up my own reference files, duplicating, etc. >Assuming Santa has limited funds, I would urge him to put them >only into programs and situations where interactive searches are >possible. How does the net feel? I agree. Mr. Hybl was not ambitious enough in compiling his wish list. An electronic library should be built around the capabilities of the available high-speed networks. I'd do this by finding or inventing an Internet protocol that users connected to the NSFnet could use to make queries of remote database server machines. The user would invoke an interactive client program on the local machine. This client would send requests to the server, which would search the database using the chosen patterns and return the results. Once the user has found interesting information, it might be downloaded directly to the screen or to a file, depending on the amount and organization of the data. It might be desirable to provide hooks by which users without NSFnet access might use the servers in batch mode, presumably by downloading the indexes by anonymous UUCP or by using a LISTSERV-style interface. A USENET-style interface would be of limited utility, unless several servers are to share all data submitted. For the end user, news readers aren't good enough at searching databases. The server model exemplified by NNTP might be a good inspiration for the proposed project, but the protocol would be different. Chuck Karish karish@mindcraft.com (415) 323-9000 karish@forel.stanford.edu
roy@phri.nyu.edu (Roy Smith) (12/20/89)
In <617@mbph.UUCP> hybl@mbph.UUCP (Albert Hybl) writes: > Speaking for myself, I think a non-interactive search would be virtually > useless [...] Assuming Santa has limited funds, I would urge him to put them > only into programs and situations where interactive searches are possible. I absolutely agree. This is the old batch processing vs. time sharing argument all over again. Batch lost out a decade or two ago, why should we bring it back in the guise of bibliographic search systems? -- Roy Smith, Public Health Research Institute 455 First Avenue, New York, NY 10016 roy@alanine.phri.nyu.edu -OR- {att,philabs,cmcl2,rutgers,hombre}!phri!roy "My karma ran over my dogma"
hybl@mbph.UUCP (Albert Hybl Dept of Biophysics SM) (12/21/89)
In message <1989Dec20.011113.724@agate.berkeley.edu> from sachs@tanner.berkeley.edu, Rainer Sachs writes: >>In article <617@mbph.UUCP> hybl@mbph.UUCP (Albert Hybl >> Dept of Biophysics SM) writes: >>There appears to be a need to establish electronic research libraries. >>... >>A UNIX shell program like lookbib or seekbib can be used to extract >>relevant reference from an M. Lesk archive. Because it is unlikely >>that all users will be able to ftp directly to the computer system >>that contains the archive, they should be able to issue a control >>command that would cause the remote execution of a seekbib shell >>with the results of the search returned via e-mail. ... > >Is this advocating non-interactive searches? [...] >Assuming Santa has limited funds, I would urge him to put them >only into programs and situations where interactive searches are >possible. I agree that electronic research libraries must be available to users on an interactive basis. However, there are circumstances where interactive access isn't necessary. For example, let's say that I may not want to take time to read certain info.groups on a daily basis. I might prefer to create an entry in my cron list that would periodically post seekbib control commands that would locate citations posted during the previous week or month that satisfy my interest profile and e-mail them to me. In an article titled "Specificity pockets for the side chains of peptide antigens in HLA-Aw68" by Garret, Saper, Bjorkman, Strominger & Wiley in Nature 342(89Dec7): 692-696, I read: "Coordinates of HLA-Aw68 have been deposited in the Brookhaven Protein Data Bank (Accession no 2HLA)." Why deny me a control initiated "seekbib PDB.2HLA" to find and deliver the coordinate file during odd hours? For your information, seekbib is a 115 line shell that I use interactively; it is my modification of lookbib. It would be a trivial matter to further modify it to produce a version that respond to remote execution requests submitted by a posted control command. I expect that an electronic library would offer several alternatives for reading or retrieving information from its holdings; seekbib could be just one choice. Sincerely yours, ---------------------------------------------------------------------- Albert Hybl, PhD. UUCP Office: uunet!mimsy!mbph!hybl Department of Biophysics Home: uunet!mimsy!mbph!hybl!ah University of Maryland Bitnet: hybl@umbc1 CoSy: ahybl School of Medicine Phone Office: (301) 328-7940 Baltimore, MD 21201 Home: (301) 243-1710 ---------------------------------------------------------------------- Responders--DO NOT USE: hybl@cs.umd.edu or ah@cs.umd.edu
hybl@mbph.UUCP (Albert Hybl Dept of Biophysics SM) (12/21/89)
The following e-mail message supports my assertion that creating electronic research libraries is technically feasible: >The fundamental problem is obtaining this information in electronic >form. Everything else is easy. ---rick Some of the information already exists in electronic form; Inertia, Policy, Money, Politics are the real obstacles. The Health Science Library (HSL) at UMAB provides a mini Med-Line service called MaryMed. At present the MaryMED data base is updated at most once or twice a month from magnetic tapes obtained from NLM. The following is a summary: } The MaryMED database contains citations to English language } articles in journals owned by the [HSL] Library. } } Contents of the MaryMED database: [Wed Dec 20 15:32:53 EST 1989] } JAN-DEC 1989 Updates 228,699 citations } JAN-DEC 1988 Updates 205,489 citations } NOV-DEC 1987 Updates 33,164 citations } ------- } 467,352 citations MaryMED averages roughly 600 citations a day while usenet generates about 2000 messages. The average MaryMED citation is about 1,500 bytes; the average usenet message is 2,358 bytes long. Usenet moves over 4 megabytes per day while MaryMED processes less than 1 megabyte per day. The usenet distributes messages among various branches and twigs of its newsgroups tree structure. The NLM likewise allows searches based on a tree structure of Medical Subject Headings. The USENET allows cross posting of a news article. MaryMED articles have many keywords associated with the reference that would be amenable for cross posting. MaryMed citations are limited to the HSL journals published in English and abstracted during the past two years. I would prefer a more complete search. The restriction to the current two years is both too long and too short. It is too short when initiating a search on a new subject where it is desired to locate reference 10, 20 or more years old. It is too short for the user wanting to search only the most recent month or two. MaryMed is a giant step forward for the HSL but only a small step toward an electronic research library. ---------------------------------------------------------------------- Albert Hybl, PhD. UUCP Office: uunet!mimsy!mbph!hybl Department of Biophysics Home: uunet!mimsy!mbph!hybl!ah University of Maryland Bitnet: hybl@umbc1 CoSy: ahybl School of Medicine Phone Office: (301) 328-7940 Baltimore, MD 21201 Home: (301) 243-1710 ---------------------------------------------------------------------- Responders--DO NOT USE: hybl@cs.umd.edu or ah@cs.umd.edu