Footnotes

...Zoo

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...reading.

I will use the terms reading and creative reading interchangeably in this presentation. I do not believe that two separate reading abilities exist; one for mundane texts and one for novel ones. Instead, my research has supported the hypothesis that a single reading ability exists. At times, it appears mundane; at other times, it can appear quite creative. The term creative reading is intended to capture this and should be interpreted as ``reading, which happens to be a creative act.''

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...experiences.

Another, perhaps more common ``reason'' for explaining why people read for pleasure is the emotional state the act produces in the reader; i.e., humans read for pleasure for the very fact that it is an enjoyable activity. This is certainly a valid reason; however, it is not one which can be operationalized given the current level of research on emotional response. Thus, I have chosen to concentrate on the ``virtual case'' approach, which is operationalizable. And, I would be willing to argue that much of the enjoyment comes from the virtual cases which are being experienced.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...Lycanthrope

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...today.

For example, Huey was one of the first to use eye movement data as a guide to his work on reading. He also discussed issues of parallel processing of a text and automatization of certain skills as readers moved from novice to expert.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...whole.

This combination of memory forms was also a feature of later reading systems, but at the time of DMAP it was a fairly novel concept.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...considered.

This has the additional benefit of allowing ad hoc categories ([#!psych:barsalou1!#]) to be represented. Certain pre-existing categories are assumed to be in the mind of a reasoner; for example, animals. However, reasoners also have the ability to seemingly dynamically create groups of concepts as needed; for instance, objects smaller than a breadbox. The ability to perform retrieval from the knowledge hierarchy based on a particular focus of the function being considered will allow this dynamic group creation to occur.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...framework.

This is a prediction of the theory resulting from the model instantiation. It should be possible to design experiments in which participants would be given stories which have been manipulated to impair various supertasks. The quality of comprehension achieved through the other supertasks should be largely unimpaired, although the theory predicts more time may be required to achieve that level of comprehension.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...preferred.

My research does not concentrate on pronominal reference; the algorithms I make use of in the ISAAC system are somewhat limited. However, stories usually contain multiple elements of confirming evidence, thereby allowing my simple algorithm to work well.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...comprehended.

Anecdotal evidence from a high school English teacher suggests that disbelief suspension is one way in which a reader gauges their comprehension level at the meta-level. More suspension suggests that the story being read is not ``making sense'' to the reasoner; as this research has shown, there are periods which require the suspension of disbelief but eventually a new set of concepts exist which no longer require the suspension. If this new set was to never be created, then the need for the suspension would never abate and comprehension would never occur. This particular teacher has had many students who did not like to read science fiction stories in the class; the reason given was that the stories were full of nonsensical ideas and there was no way to comprehend the stories.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...impossible.

For those interested, a tocsin is an alarm and a baloo is a bear. These are English words which have fallen out of use. Mirar is the Spanish verb to see; I added English verb endings to the foreign word. Foozle and calaporatory are imaginary words meaning to raise, to sound and constructed out of cardboard, respectively. Thus, a tocsin is an alarm which you would want to raise upon seeing a bear, unless the bear is made out of cardboard.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...provided.

This may seem counter-intuitive; to most people, this example would make more ``sense'' if interpreted as John is bear-like, a metaphorical usage. Remember, however, that these ontological constraints have the most effect when no other information is known. This information can be background knowledge or story knowledge, so a reasoner already familiar with the bear metaphor may retrieve that interpretation instead of this default one. It is important to note that my approach handles metaphor as a normal part of the understanding cycle. Since metaphors are pervasive in language (see [#!metaphor:carbonell1!#,#!phil:johnson1!#,#!metaphor:lakoff-johnson-1980!#], for example), I consider this ``unified'' handling of non-metaphors and metaphors to be an important feature of my overall theory.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...needed.

As stated earlier, this supertask could also be used to comprehend real-world scenarios as well as literary ones, if the reasoning system was significantly expanded to need such comprehending skills.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...Different.

Read in this case means to consult a disk file line by line in order to simulate the reading process. While I do not expressly model eye-movement, as in Just and Carpenter ([#!read:just!#]), ISAAC does have the capability to skip around in the file in a nonlinear fashion. This approximates a human reader's ability to reread sections of the text, if needed. For example, if ISAAC is skimming a text and suffers a comprehension failure, the first action which would be taken would be to reread the last few sentences which had been skimmed.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...aspect.

As will be pointed out in the next chapter, a number of evaluation questions dealing with Zoo asked the reader to explain the irony of the title. ISAAC's limited literary knowledge could not provide a completely satisfactory answer, but this dual concept of zoo is close to the irony represented in the title.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...reader.

The term capable in this context means that ISAAC was judged at the same level of competence as a group of humans on three stories. This judgment resulted from independent evaluators. Also, a capable reader in this case is restricted to mean a capable, high-level reader. As I have pointed out earlier, ISAAC does not attempt to model certain low-level reading behavior, such as eye-movement. So, it is technically incorrect to say that ISAAC is ``as capable of a reader as a human.'' These caveats should be kept in mind as ISAAC's performance is further analyzed.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...for.

The irony and similar questions fell into the category of incorporation questions, which the raw results show ISAAC as performing slightly worse on than the human participants; see Table 5.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...be.

A similar problem arises in research involved with the formation of novel concepts (see, e.g., [#!concept:shoben-1993!#,#!creat:ward1!#,#!psych:wisniewski1!#]). If a novel concept is formed using existing concepts, what category should the new concept belong to? For example, if pet and fish are combined, where in a knowledge hierarchy should the resulting concept be placed?

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

Kenneth Moorman
11/4/1997