Introdsscfng Sci-A4ate- A Menu. IMven Microcomputer Software Package for Onffne assd Offlfsw Information Retrfeval. Part 1. The Sci-Mate Personal Data Manager .— ..- —- Number12 March 21, 1983

In 1982, about three million micro- In either case, the next problem is how computers were delivered in the US to file them. alone. I The average number of users While there are many -organized easily exceeds double that number. reprint collections, most scientists lose Among them are thousands of scientists their initial enthusiasm for maintaining and scholars who are Current Contents@ them. They start out with the best inten- (CC”) readers. By the end of this tions, but soon abandon any attempt to decade, I estimate that over 90 percent be systematic. PRIMATE was designed of our readers will rely on microcomput- to help scientists keep up with their ers or intelligent terminals to store and reprint files by eliminating the indexing retrieve information. and filing problems’ in the ordinary sense The first time I wrote about ISI@’s of those terms, plan to produce software for microcom- The enthusiastic response to the idea puters we received several hundred let- of a personal reprint management sys- ters—more than on any other subject tem encouraged us to continue the de- I’ve ever dkcussed in CC. The micro- velopment of PRIMATE. PRIMATE I computer system for managing reprint and PRIMA TE H represented the early files I described was designed by A.E. stages of this system. o We have since de- Cawkell, ISI’S former dkector of re- cided to name the system for the main search. Cawkell continues as a consul- audience it will serve-the scientific tant to 1S1. He also publishes a newslet- community. So we changed the name to ter, Information Technology and People Sci-Mate ‘“ to eliminate the implication (CITECH Ltd., P.O. Box 5, Ickenham, that it serves the exclusive needs of those Uxbridge, Middlesex UB1O 8AF, in primate research! It will also serve to England). highlight the fact that the new Sci-Ma te Let’s recapitulate the reprint filing system is significantly advanced over its problem and the solution first proposed predecessors and can be used in con- by PRIMATE, that is, Personal Retriev- junction with MI’s scientific and schol- al of Information by Microcomputer arly data bases. And Terminal Ensemble. Scientists ac- The original system was designed al- cumulate reprints through various most exclusively to deal with the prob- means—writing to colleagues, using re- lem of filing and retrieving personal re- print cards such as Request-A -Printa, print collections in an offline situation. photocopying journals in the library, or Thk dktinction from online means that by ordering tear sheets from 1S1. When it does not have to be operated through a the reprints arrive, you peruse most of centralized time-shared computer sys- them quickly, while a selected group is tem. Each microcomputer has its own read carefully or discussed in your lab. central processing unit (CPU). We are

80 also developing a version of Sci-Mate The Sri-Mate Personal Data Manager which can be used in minicomputer is a data base management system de- systems that serve many terminals in a signed specifically to accommodate tex- single institution. In that situation each tuaf material-especially bibliographic terminal is used in an online mode in re- material. You can, of course, use it to lation to its CPU. store full bibliographic citations. How- In its four-year evolution to Sci-Mate, ever, it can also handle notes and the 1S1 system has become much more abstracts, or comments that you wish to than a reprint fiting system. First, it is a add to help you retrieve them or to personal offline data base management W3htate or sort them once retrieved. system with unique features and capabil- Sci-Mate’s forerunners were designed ities not available in a single package Imainly with reprint coUection manage- elsewhere. Second, it is a system for ac- ment in mind. Sci-Mate itseff has re- cessing and using numerous commercial tained that characteristic. In fact, the online time-shared data bases that does ability to manage reprints may be the not require knowledge of their special function that will be initially the most command languages. The offline and useful to most CC readers. However, its on[ine components are available sepa- usefulness in other record keeping situa- rately or in combination. In a future es- tions wilf become apparent here. say, I’ll d~cuss Sci-Mate’s online capa- Let me make clear that in the follow- bilhies in detail. In this portion of the es- ing discussion the machine readable bib- say, I’ll focus on its offline functions. liographic records are only representa- Figure 1 summarizes these features and tions of the original reprints, letters, or their benefits. other documents. The actual reprints A microcomputer data base manage- are filed separately. The same would be ment system is a software package, that true for correspondence, lab notes, and is, a series of integrated computer pro- medical case reports, unless the fuff text grams which enable you to create and is so short that it constitutes an entire manipulate many types of records. Most document. of the commercially available software To use Sci-Mate as a reprint file packages were designed for use in busi- manager, you have several options. You ness settings. They are commonly used could assign each new reprint a serial to store lists of customers, subscription number. For each reprint, you’d enter records, inventories, etc. They are de- this accession number and a complete signed mainly to handle numeric infor- bibliographic description or summary. mation, and are limited in their flexibil- You could do thk one reprint at a time ityy for handling textual information. or you could have an assistant add the in-

Features Benefits

Random access storage. Fast access to any stored record. Menu-driven command system. No need to learn a special language. Tutorial subsystem. Sci-Mate lists options for you. Variable length records, You can store textual data of any length. No pre-indexing. You can search any phrase, word, or character alone or in combination. Flexible record layout. You can design your own record format. Report generator. Columnar reports to your specifications. Immediate record/file updates. Records searchable as soon as entered. Eliminates need to create or update lists.

Figure 1: Features and benefits of the Sci-Mate ‘“ Personal Data Manager

81 WELCOME TO THE 1S1 SCI-MATE SYSTEM

Do you want to:

1, Enter the UNIVERSAL ONLINE SEARCHER 2. Enter the PERSONAL DATA MANAGER 3. Enter or change date 4. Delete the WORK file 5. Receive instructions on the use of SCI-MATE 6. Leave the SCI-MATE system

Select a number or enter ? for help:

Ftgure 2: This is what you see upon entering Sci-Mute m. formation for a series or batch of papers, Having stored several hundred docu- perhaps after reading CC or as reprints ment descriptions, how do you retrieve arrive each week. them? Sci-Mate is designed to be “user- However, with Sci-Mate you don’t -friendly,” which is computerese for have to assign a number “manually.” “easy to use.” The validity of such a You can use the number that the pro- claim depends upon the amount of artifk gram automatically assigns to each en- cial intelligence built into the system. try. In Part 2 of this essay I’ll describe We believe Sci-Mate is justifiably called how much of this keyboard work can be user-friendly because it is a “menu- eliinated if you choose to input records driven” or multiple-choice system. At by using the online option to “capture” each step the computer asks you a ques- records. tion and presents a short list of options. As the information for each paper For each choice you press one key to is keyboarded, Sci-Mate automatically execute the option or command chosen. adds the new records to the computer’s Figure 2 shows the first Sci-Mate ques- memory. The reprints themselves can be tion after you insert the magnetic floppy fried in numerical order in filing cabinets disk (diskette) on which the program is or loose-leaf binders. If you prefer, in- stored. You are presented with six stead of fiiing numerically, you can file choices. Figure 3 shows what happens the papers alphabetically—either by when you press “2”, the request to author or by subject categories. Howev- search your personal data base. Again er, the serial number system maintains you have six options, one of which sim- your file esaentiaUy in chronological ply takes you back to the beginning. The order, This is handy when you want to first option is the one you can expect to browse through the reprints you ac- use a lot for searching. The second op- quired over a particular time period. tion is used when you want to enter new However, if you prefer to file the actual reprints into the collection. Figure 4 reprints by lecture topics, you can in- shows what you see when you press clude the name or code for the file folder ,4 1!, —“Search the USER file.” as part of the document description. In- Figure 5 shows what you see when you cidentally, as you are keyboarding all press “2”. You enter a search term, th~ information, or capturing it from an- which is used to search the entire file. other fde, it is displayed on the screen This means you do not have to specify for visual verification. whether you want to search titles, author

82 OFFLINE Sci-Mate also includes a way to ------broaden your search. It gives you the op- You may: tion of “truncating” your term. Suppose you wanted to review the papers in your 1. Search the USER ffle cofJection which contained the term 2. Enter a new record 3. Display/copy WORK file records bronchitis. You can broaden the search 4. Create and update tenrpfates by inserting a “, ” represented in 5. Genemte document request Sci-Mate by this symbol: #. So if you status report 6. Return to the SCI-MATE menu entered “bronctdl”, you would get hits using the terms bronchial, bronchial-se- Select a number: cretions, bronchoconstn”ction, broncho- scopic, and so on. You can afso put the “wild card” at the front of a term to fmd Ffgmre 3: This is what you see after selecting option prefixes to the key word. You can even 2 in Figure 2. insert it in the middle of a term. Almost all data base management sys- names, journals, etc. Everythktg in the tems claim to “search” files. Even a word records you have entered is automatical- processor can do limited searching. But ly searched. If you had added your own it is absurd to discuss the search capabih annotations or comments to your files, ty without specifying the time required those comments would also be searched. to do the average search. We believe This “string searching” capability is a that Sci-Mate is extremely fast for large, simplifying element in Sci-Ma te that is randomly organized fries. But th~ state- presently available only on large text ment is best demonstrated on your own searching computers. or comparable files. In any case, the One feature I should stress is that speed is affected by many factors. every entry or change you make is imrne- There is a significant difference de- cfiately searchable. No matter how many pending on whether you are storing data times a day you add or modify records, on a hard rather than a fIoppy disk. The you can search them as soon as you a vemge search rate for the much larger create or change them. hard disk is about 100 records a second, Sci.Mate also allows you to refine or given an average of 300 characters per extend your search through use of record. For floppies, the speed is about Boolean logic. Thk means you can com- one-thwd of this. bine terms. For example, you could If you are used to working with a search for papers on fission. But in a less mainframe or minicomputer system, it is restricted search both fusion and fission easy to make invidious comparisons be- could be stipulated. This in turn could be narrowed to a particular journal or laboratory. SEARCH Perhaps the most common biblio- graphic search involves the location of a Would yOU hke to: paper by a known author on a particular 1. Search by accession number topic. All you remember is one keyword 2. Search text: find total hits before of the title. So you simply use the and dispfrly command (or “operator”) to find the 3. Search text: display hits as retrieved 4. Return to the OFFLfNE menu Jones paper on mitochondria. Or you remember Smith wrote a paper in Scien- Select a numbec tific American a few years ago, but you simply can’t recall the title. Just key in Figure 4: Ths iswhatyousee sfter selectingoption the author’s name and the journal name. 1 in F@re 3.

83 of a short manuscnpt, e.g., a letter to the TEXT SEARCH ------editor of Nature or Science, as a single record. Furthermore, you could in prin- Enter a search sia!ement ciple store sections of a large manuscript

computer as a series of “linked” separate records. It should be obvious, however, that

Figure 5: What you see when you ask Sci-Mute ‘Mto when you use Sci-Mate for such lengthy search by text. The user has entered “’computer. ” records you sacrifice your ability to store The search term will be used to “string search many smaller records, However, you the entire record. could also use a hybrid system comb~n- ing storage on hard disks for the basic tween Sci-Mate and software designed reprint file and use floppy disks for one to operate on such relatively large time- or more complete article texts. shared computers. Most large-scale sys- The Sci-Mate Personal Data Manager tems use an inverted file system. Sci- not only allows you to store the biblio- Mate has eliminated the need to create graphic data for your reprint requests, inverted files and the updating problems but also review their status periodically. associated with them. But thk means it Suppose, for example, you want a paper will be used optimally for files that are you saw in CC or identfled while using typical of a personal reprint collection. the Sri- Mate Universa[ Oniine Searcher. Do not expect to store a year of CC or You order a reprint before you add the Science Citation Index@ on Sci-Mate. In reference to your permanent data base. test fdes of 5,000 records containing an Then you enter this information in your average of 300 characters per record, a data base, which contains all your recent typical search question involved less requests and other items. You turn on a than one minute of search time. Inciden- “flag” or code to be reminded that tally, the search time goes down as your you’ve sent out a reprint request card or request becomes more specific. wrote the author. When the reprint ar- Sci-Mate is geared up for a large rives, the flag can be turned off. An ex- number of bibliographic or other texts ample of a flag screen is shown in Figure such as correspondence or even patient 6. You can use it to be reminded about records or lab notes. The number of rec- up to six different things: OATS@ ords you can store depends on how (Original Article Text Service), interli- lengthy those records are, and what your brary loan, Lockheeds DIALORDER, disk capacity is. The maximum length of Request-A -Print, telephone calls, etc. a separate record is 1,900 characters. Subsequently you can review your Even at that size you can store 2,200 records to obtain a printout of all records on a 5-megabyte hard dkk de- vice. There are two such devices on a 10-megabyte disk. ~ STATUS However, most people use records of less than 1$3(K!characters. In thk case, 1. OATS ON 2. ILL ON you can store up to 6,(XH3records if they 3. REP. REQ. OFF are only 60Q characters long. The typical 4, DIALORD. OFF OFF journal article title usually consists of 5, SCHED. 6. RESPOND ON fewer than 100 characters. To exceed the storage limit of the hard dkks you Would you lie to changea flagstatus (Y/N) ? can use a series of floppy disks or an ad- ditional hard disk. Ffgure 6 An example of a Sci-Mate ‘“ “fSag,” to In principle there is nothing that pre- keep you up to date on the status of document vents you from storing the complete text delivety.

84 Field Field A sample template is shown in PIgure No. Name No. Name 7. Like all Sci-Mate templates, it consists of up to 20 fields. In this example, fields 1 . No. 11 Compound 2 Title 12 LD50 one through nine have been taken up 3 Language 13 Lab with the accession number, title, origi- 4 Dec. Tp. 14 nal language, document type, author, 5 Author 15 6 Corp. so, 16 corporate (or organization) source, 7 Journal 17 journal, year, and OATS order number. 8 Year 18 Fields ten to 13 have been taken up by 9 OATS # 19 10 Org Test 20 organism tested, test compound, LD50 (lethal dose, 50 percent) test results, and Ffgmra 7: An example of bow you might arrange a laboratory. (The LD50 test measures the Sci-Ma/e ‘“ “template” to help you sort and dose of a substance that kills haff the test search. animals. ) Although the field names are limited to eight characters, there is noth- “flagged” items. This status report will ing to prevent you from putting the fulf remind you that you’ve not received, for titles, authors’ names, and so on, into example, the reprints or interlibrary loan the appropriate fields. Fields 14 to 20, in items you requested. This type of follow- this example, have been left . They up system also can be used as your “tick- can be filled with other items at a later ler” file to check on correspondence that time. You might, for instance, want to is overdue. Flag 6 of Figure 6 shows how add a field for “notes,” and enter your you can indicate that a letter needs to be opinion of the data in the record. You answered. I’m using Sci-Mate for this can arrange fields in any order you wish, purpose and, more importantly, as a way and you can use as many as 50 dtiferent of checking on the status of small and template formats. large writing or research projects. These templates enable you to sort the There is a feature of Sci-Mate that records obtained from a search by se- causes quite a stir wherever it is demon- lected fields. If you choose to sort the strated. It is the menu-driven system for “author” field (field 5 in Figure 7), the formatting, or “templating,” your files. result is a display of records, shown in al- A template is simply the sequence of in- phabetical order by author. You might, formational elements for each record. In however, want to sort by compound. So computerese, these elements are called you would order the system to sort by “fields,” and should not be confused field 11. The results would then be a with fields of science or research. Using display of records in alphabetical order templates will help you maintain consis- by compound. (The sorting program au- tency in indexing and storing records. tomatically alphabetizes. ) You can also They also label the records you dkplay, sort by language, document type, jour- and are used to sort the results of nal, organism tested, laboratory, notes, searches. or any other field.

ACC. NO. ORGANISM COMPOUND LD50

105 Rats Caffeine 192 mg/kg 283 Rabbits Fenoflurazole 28 mg/kg 1205 Rabbits PhenylbutaTone 146 mg/kg 1299 Mice Carbofuran 2 mg/kg

Figrrra8: An example of a repml generated by Sci-hfafe ‘u.

85 With Sci-Mate you can view hits on Nothing can replace adequate docu- the screen, or use a feature called the mentation and service. Each Sci-Ma te “report generator” to print lists of customer receives a detaifed manual. 1S1 specfled information. This feature en- has also set up a special customer hot ables you to print lists of fields of a line so that unanticipated problems can record in a horizontal rather than a verti- be dealt with quickly. Don’t laugh— cal format. Suppose you wanted to re- some scientists forget to plug the cord port on the organisms you’ve tested, plus into the electrical outlet. Beyond the ob- the compounds you used and the LDSO vious, those problems that cannot be test results. Sci-Mate can give you sim- dealt with over the phone wilf be re- ple step-by-step instructions to produce ferred immediately to our research and a report like the one in Figure 8. The first development department where they column shows the record accession will be dealt with by the team that has number. The second lists the organism; helped develop .Sci-Mate. We cannot be the third, the test compound; the fourth, responsible for any equipment failures the result. This figure is arranged by ac- but we can attempt to tell you if you are cession number order. But you could having a hardware rather than a software generate th~ report in alphabetical or- problem. And quite frequently it maybe der by any field, using the sort facility. a matter of improper execution of your You can also generate” lists using other own specifications for format, etc. Using fields, and get reports on, say, the au- a new system to organize your files is a thors in your file, the status of projects, significant learning experience. journals, and so on. The 1S1 staff is excited about the p~ The microcomputer software market tential of Sci-Mate. We think it’s a logi- is an enormous and growing industry. cal step in our mission to improve access One indicator of this is the number of to scientflc information. When we start- magazines catering to the microcomput- ed this project four yearn ago, we er buff, such as Byte, Personal Comput- planned to build our own special pur- ing, and Info World. One of the most im- pose microcomputer. Now they are so portant functions performed by these cheap and diversified that it is neither publications is evaluation of software necessary nor practical for us to manu- packages. One of the prevalent criti- facture or assemble hardware. But Sci. cisms is that software is not adequately Mate may well be the first reason you tested on different microcomputers. may acquire a microcomputer. Whether Since software packages are sold at bar- you acquire anew microcomputer or use gain prices, suppliers will not back up the one you have, do not expect Sci- their software with “service ,“ But all of Mate to make it the ultimate intelligent Sci-Mate’s functions have been tested by system. As with any decent software de- several outside groups. And 1S1 will velopment program, we are making pr~ stand behind its software claims. visions for continuing improvements However, it is not enough to say that and updates to the software. This will be our software will be backed up by ISI’s particularly relevant for the online com- reputation as a user-sensitive company. ponent to be described in the second Whiie Sci-Mate has been tested in part of thw essay. Updates in our soft- numerous commercial and academic in- ware will include such features as auto- stitutions, there is no way that one can matic re-formatting of citations depend- anticipate alf the myriad problems that ing upon the journal in which you pub- can arise when a system is used by thou- lish. sands of individuals, We have taken con- .Sci-Mate software is currently avail- crete actions to ensure that the software able for the IBM Personal Computer, works to your satisfaction. Vector 3 or 4, the Apple II, the TRS-80

86 Model II, and any CP/M-80 (control system is not a toy. It requires 64K program for microprocessors) system RAM. The machines required are in the with a standard eight-inch disk. This in- price range of S4,000 to $7,500 in the cludes dozens of brands lie Altos, In- US. These prices include all the neces- dustrial Microsystems 8000, and Hew- sary peripherals. lett-Packard 125 (8-In.). For the IBM, a The offlie Sci-Mate Personal Data CP/M-86 operating system is required. Manager costs S540. The online Sci- For the Apple II, you need cards (circuit Mate Universal Online Searcher option boards) for a Z80 microprocessor, 16K is $440. Together, both cost $880-a RAM (random access memory), and an savings of $100. For more information 80-column screen. Of course, a printer is on Sci-Mate, contact Joel Sartorius, 1S1, also needed. In addition, to use the Sci- University City Science Center, 3501 Mate Universal O;line Searcher, you Market Street, Phdadelphla, Penn- need a modem, a device you connect to sylvania 19104, USA. Telephone: your telephone so you can dial up large (800)523-4092. Telex: 34-5305. AILothers online data bases. We also recommend should write to ISI’S US headquarters. 500 kilobytes of dkk storage for an effec- In the near future, I’ll describe the tive system. Sci-Mate requires at least Sci-Mate Universal Online Searcher, two disk devices. It can also operate with which can be used for online searching, two floppy disks as a minimum. How- and how the two Sci-Mate systems work ever, it’s preferable that one of your in tandem. The online system not only d~ks be a hard one, to store your rec- diaJs up the host computer automatical- ords. Hard disks store much more infor- ly but also eliminates the need to learn mation than floppies. command languages. As they say in the It is inevitable that readers who do not US. “Look ma—no hands.” own a microcomputer will ask us to rec- ommend the best. That is not a simple question because each microcomputer will perform better under certain condi- tions. However, within the next few ***** months we will be evaluating the results of Sci-Mate use in the field and will be glad to pass along to you our experiences and recommendations. It will be obvious to experienced mi- My thanks to Thomas Marcinko for crocomputer users that the Sci-Mate his help in the prepamtion of this essay. @t% 19

REFERENCES

1. Frfedrfch O. The computer moves in. Time 121(1):14-24, 1983. 2, Guffeld E. Introducing PRfMATE-Personal RetrievaI of Information by Microcomputer A nd Terminal .Ensemble. Essays of an information scienti$t. Philadelphia: 1S1 Press, 1980. Vol. 3. p. 551-5. (Reprinted from: Current Content.! (29):5-9, 17 July 1978.) 3. ------On the 25th anniversary of Currenf Contents/L@ Sciences we look forward to the electronic online microcomputing era. Current Contents (4):5-11, 25 January 1982.

87