LISP LORE: A Guide to Programming the LISP Machine, (2 °~ Edition) Hank Bromley and Richard Lamson Kluwer Academic Publishers $47.50 ISBN: 0 - 8 9 8 3 8 - 2 2 8 - 9 R e v i e w e d by: Keith Price Considering my reaction tc the first edition of LISP LORE: A Guide to P r o g r a m m i n g the LISP Machine (see Sigart #100, I generally liked it, but the o p e r a t i n g system changed), I was pleased to see a second. This new edition is based on the first with substantial additions and r e w r i t ing (by Richard Lamson of Symbolics) to account for the changes made by S y m b o l i c s between version 6.1 and 7.0 of their system. The book is s o m e w h a t expanded and rearranged (flavors are no longer the first topic that is introduced) with n e w e x a m p l e p r o g r a m s (as before, source is also available on tape). LISP machines (from any of the f e w manufacturers) present a challange to a begining user since the style of effective usage is different from most o t h e r operating s y s tems (e.g. Unix, VMS, etc.). Even t h o u g h c o m m a n d s and features differ, the basic initial usage of the c o m m o n interactive systems (and even the old card based systems) is very much the same. The LISP machine families seem to require more initial i n f o r m a t i o n before any usage is possible. This hurdle t e n d s to eliminate the casual user there is a large i n v e s t m e n t of t i m e to k n o w enough to do simple things so it is either used for e v e r y t h i n g or n o t h ing. But once past this initial step, the ease of use and t h e overall system capabilities make such systems e n j o y able to use. As a (2-year+) user of a LISP machine, does this b o o k teach me anything? Probably not, e x c e p t for small things that become clearer, or n e w f u n c t i o n s and features that are mentioned. But much the same thing happens any t i m e I read the manuals; I find some n e w useful feature that I probably was n o t looking for and never knew I needed. Do new users find t h e b o o k helpful? Based on h o w the first version was used by some of t h e other users in o u r group, I think the i n t r o d u c t i o n is short enough for n e w users to get past t h e initial shock of the 13 v o l u m e manual set and b e t t e r organized than the i n t r o d u c t i o n in t h e manuals. The a d v a n c e d features such as w i n d o w s , graphics, files and b u i l d i n g systems are described well e n o u g h for a n e w user to g e t started and to use these features effectively. It is then easier to read the manuals to learn details after y o u are already using simple v e r s i o n s of these c o m p l e x s u b s y s t e m s . The book fails to a n s w e r all the o b s c u r e q u e s t i o n s that arise through using such a system. (For example, w h a t causes p r o g r a m s to g e t an error w h e n adding t w o bignums, but not always the same b i g n u m s , or the same place, but repeatabty with a given pair of b i g n u m s deep in a given program? After several days of messages, the only guess was hardware. One system was "cured," and another has had a similar problem in different programs.) It also does not address another recent (and recurring) problem where we had a c o m p l e t e seizure of all the s y s tems upon a t t e m p t i n g any n e t w o r k access (the solution required disconnecting taps from the physical n e t w o r k u n til the p r o b l e m was found, the failure affected every s y s tem on the cable and through all the bridges), but the d i s cussion of networks, processes, and scheduling at least helped understand w h y this problem occurred. Overall I like this edition better than the first and e x pect that it will be used, both by current users of our s y s tems and by new users over the next f e w years. MACHINE LEARNING: A Guide to Current Research T. M. Mitchell, J. G. Carbonell, R. S. Michalski Kluwer Academic Publishers $55.00 ISBN: 0 - 8 9 8 3 8 - 2 1 4 - 9 Reviewed by: Keith Price Machine Learning: A Guide to Current Research p r o v i d e s a collection of current research papers m o s t l y collected form participants at the Third International Machine Learning Workshop held in 1985. These 77 separate papers were collected and reformatted at Rutgers to produce a c o n s i s t e n t format and a c o m m o n b i b l i o g raphy. The range of learning topics include (taken from the preface): analogy, conceptual clustering, e x p l a n a t i o n based generalization, incremental learning, inductive i n ference, learning apprentice systems, machine discovery, theoretical models, and applications. All of these topics, except incremental and apprentice s y s t e m s are included as heading in the index. Other useful index headings include research institution (a few are missing, primarily t h o s e where the several authors are from m o r e than one place) and p r o g r a m name. The papers are short; the average length of five pages per paper, based on the n u m b e r of papers and the length of the book, o v e r s t a t e s their length because of blank pages due to f o r m a t t i n g considerations. In the short space for each paper it is not possible to e x plain in detail h o w a n y t h i n g is done. There is space only to say w h a t is being done by the system. Thus, the bibliographic references b e c o m e i m p o r t a n t for a deeper understanding of individual techniques. As a researcher in a field o t h e r than machine learning, I can n o t evaluate the quality of each of the papers. The three editors are b e t t e r able to make that judgment. In the preface t h e y call this book a "snapshot" of the field and indicate that it " p r o v i d e s a representative s a m p l i n g of the best o n g o i n g w o r k in the field." As such, the b o o k serves the purpose of a research r e v i e w for the entire field rather than a description of h o w to duplicate various research results (as should be d e m a n d e d of refereed j o u r nal publications, for example). This makes the b o o k a good initial reference, especially with the index and the single bibliography, both for researchers in machine learning and those outside the field w h o want to learn about it or at least keep generally aware of its major directions and concepts. Also, I see this as a research or reference book more than a text book. The reformatting of the various submissions is a m o n u m e n t a l task (this c o m e s from the experience of doing similar things for the N e w s l e t t e r for a f e w years), and the layout, etc. was generally g o o d . The reference information and style was s o m e w h a t inconsistent, most likely r e f l e c t ing what was s u b m i t t e d by the original authors. (In the Newsletter, I find m o r e f o r m a t t i n g , spelling, etc. errors in the o n - l i n e submissions or the original h a r d - c o p y than in what is r e - t y p e d here.) The book gives me a better perspective of the machine learning field and may be helpful for my l o n g - t e r m research since learning is ult i m a t e l y the m o s t i m p r o t a n t feature of an AI system. SIGART N e w s l e t t e r , J a n u a r y 1988, N u m b e r 103 Page 22

/lp/association-for-computing-machinery/book-review-machine-learning-a-guide-to-current-research-by-t-m-p0uk0Cmc1Z