Hartley A., & Paris C., (1996) Two sources of control over the generation of software instructions. Information Technology Institute Technical Report Series, ITRI-96-02. Also published in Proc. ACL, Santa Cruz, California, June 1996. (pp 192---199).  Home/Search   Document Details and Download   Summary   Related Articles   Check

....of previous works on lexical and syntactical closure of sublanguages (e.g. Kittredge, 1987) a small corpus appeared sufficient. The corpus (consisting of a set of parallel texts from the CAD CAM domain and a set of non parallel instructional texts from other domains) was analyzed following (Hartley Paris, 1996), tagging for the type of text unit (e.g. procedure, description) plan element (e.g. goal, side effect, step) and a set of SFL lexico grammatical features. The analysis revealed the following tendencies of similarities and differences between the linguistic resources needed for each ....

Hartley A., & Paris C., (1996) Two sources of control over the generation of software instructions. Information Technology Institute Technical Report Series, ITRI-96-02. Also published in Proc. ACL, Santa Cruz, California, June 1996. (pp 192---199).

.... DRAFTER and ISOLDE s aim is to generate multilingual manuals from a unique knowledge base [27, 31, 17] Some of these Projects start from an analysis of the manuals of some well known software products, to examine the types of information they include and the linguistic structure of each of them [15]. By adopting the metaphor of emulating the ideal of having an expert on hand to answer questions , I Doc (an Intelligent Documentation production system) analyses the interactions occurring during expert consultations, to categorize the users requests and to identify the strategies they employ ....

....the users according to their experience with the application) XDM Agent s model is more interesting, since its behaviour is driven by some personality traits that we describe in terms of its helping attitude . Let s see this in more detail. A typical software manual includes three sections [15]: i) a tutorial, with exercises for new users, ii) a series of step by step instructions for the major tasks to be accomplished and (iii) a ready reference summary of commands. To follow the minimal manual principle ( the smaller the manual, the better : J M Carroll, cited [1] the Agent ....

Hartley, A. and Paris, C.Two sources of control over the generation of software instructions. Proceedings of ACL 96, 1996.

.... in a set of instructions, has also been explored [Dale, 1992] Expressing the purpose of an action, i.e. Do x to do y, has been examined by several researchers as well [Di Eugenio, 1993; Vander Linden, 1994; Kosseim and Lapalme, 1995; Vander Linden and Martin, 1995; Di Eugenio and Webber, 1996; Hartley and Paris, 1996] Despite all of this previous work, discussed in Chapter 2, the issue of expressing an action s termination remains unexplored. Before discussing how termination information is expressed, I should clarify the terms 4 which I will be using to refer to actions and their descriptions in Natural ....

....and a collection of technical orders (military instructions) for the maintenance of F 16 aircraft. It considers only the numbered step by step parts of the texts rather than the general discussion in the former and the notes, cautions, and warnings in the latter. In a corpus study done by [Hartley and Paris, 1996], such step by step instructions are shown to be a sub genre of instructions. Their analysis shows that step by step instructions have linguistic features, such as the dominance of imperative sentences, which distinguish them from other subgenres (e.g. reference and tutorial texts) Thus, ....

[Article contains additional citation context not shown here]

A. Hartley and C. Paris. Two sources of control over the generation of software instructions. In Proceedings of the 34th Annual Meeting. Association for Computational Linguistics, June 1996.

....projects (Binstead et al. 1995) Grasso, 1997) which aim to influence patients decisions on smoking cessation, and the RAGs project (Cooper at al. 1996) which assists in genetic counselling. Another area focuses on technical documentation generation, DRAFTER (Paris and VanderLinden, 1996) (Hartley and Paris, 1996), TECHDOC (Rosner and Stede, 1992) and IDAS (Reiter et al. 1995) for example. In all these cases, the written textual output is produced without intervention by a user there is often a knowledge elicitation or acquisition stage, but this entirely precedes the generation phase, which is itself ....

"Two Sources of Control over the Generation of Software Instructions" in Proceedings of the Annual Meeting of the Association for Computational Linguistics

....projects (Binstead et al. 1995) Grasso, 1997) which aim to influence patients decisions on smoking cessation, and the RAGs project (Cooper at al. 1996) which assists in genetic counselling. Another area focuses on technical documentation generation, DRAFTER (Paris and VanderLinden, 1996) (Hartley and Paris, 1996), TECHDOC (Rosner and Stede, 1992) and IDAS (Reiter et al. 1995) for example. In all these cases, the written textual output is produced without intervention by a user there is often a knowledge elicitation or acquisition stage, but this entirely precedes the generation phase, which is itself ....

"Two Sources of Control over the Generation of Software Instructions" in Proceedings of the Annual Meeting of the Association for Computational Linguistics

....where we have given a literal translation (in this type) after each example from our corpus. These translations are almost always acceptable English instructions. On the other hand, other research has shown preferences in rhetorical relations according to the language studied (Delin, Scott, and Hartley 1996). To describe the natural process of generating instructions, we have taken a corpus of texts and analyzed it both at the semantic and the rhetorical levels. When selecting the corpus, we took a few precautions: we rejected what seemed to us the obviously flawed texts and we only considered ....

....nucleus and satellite will be determined at the rhetorical level when the most appropriate relation is chosen. To define the senses, we have been inspired by the semantic relations of type generation, instance generation and enablement of Goldman (1970) which have been used by Delin, Scott, and Hartley (1996) as the basis for the generation of multilingual instructional texts. These relations, however, do not cover the whole range of senses found in instructional texts; thus, we have developed more specific entities particular to this discourse genre. The senses have been identified only through ....

[Article contains additional citation context not shown here]

Hartley, A. and C. Paris (1996, June). Two sources of control over the generation of software instructions. In Proceedings of the 34 th Annual Meeting of the Association for Computational Linguistics (ACL-96), University of California, Santa Cruz, pp. 192--199.

....and a collection of technical orders (military instructions) for the maintenance of F 16 aircraft. I have only looked at the numbered step by step parts of the texts rather than the general discussion in the former and the notes, cautions, and warnings in the latter. In the corpus study done by [Hartley and Paris, 1996], step by step instructions are recognized as a subgenre of instructions. Their analysis shows that step by step instructions have linguistic features which distinguish them from the other sub genres (e.g. reference and tutorial texts) Thus, focusing on the step by step parts of corpora is ....

.... syntactic constructions, as described by [Elhadad et al. 1997] The development of a lexical choice algorithm begins with determining the correlations between the contextual factors and the linguistic features of words and constructions, usually through a corpus analysis as demonstrated by [Hartley and Paris, 1996], among others. Once the contextual factors and the ways in which they constrain the range of linguistic features have been determined, several methods can be used to perform lexical choice. Since generation systems depend on lexical choice to determine the most appropriate way to express ....

[Article contains additional citation context not shown here]

Anthony Hartley and C'ecile Paris. Two sources of control over the generation of software instructions. In Proceedings of the 34th Annual Meeting. Association for Computational Linguistics, June 1996.

....and in addition, we took into account the textual type as a criterion for choosing grmmatical structures. Indeed, our corpus analysis revealed a strong correlation between the textual genre (execution, hybrid or comprehension texts) and the grammatical form of rethorical relations; thus confirming [Hartley and Paris, 1996] s results. Figure 10 shows the output of the grammatical structurer. The lexico morphological selector takes over the grammatical structure and selects the lexemes to be used. The words are chosen (in the current implementation, a one to one correspondence exists between concepts and words) and ....

Hartley, A. and Paris, C. (1996). Two sources of control over the generation of software instructions. In Proceedings of the 34 th Annual Meeting of the Association for Computational Linguistics, pages 192--199, University of California, Santa Cruz. ACL.

....communicative goal, thus providing the flexibility to produce several texts corresponding to the same set of instructions. This was illustrated in Figures 4 and 5. 10 We have recently completed a corpus analysis to uncover the differences in the linguistic realisations of the these two genres (Hartley and Paris, 1996). Multilingual Document Production 19 GammaEarly multilingual drafts: Because the document planning and the composition tasks are partially automated, it is possible to produce multilingual drafts for portions of the knowledge base as they are completed. GammaPropagation of changes: Because ....

Hartley, A. F. and C. L. Paris: 1996, `Two Sources of Control over the Generation of Software Instructions'. In: Proceedings of the 1996 Meeting of the Association for Computational Linguistics (ACL-96). Santa Cruz, California, USA. Also available as ITRI report ITRI-96-02.

.... this part of the system, we asked the evaluators to critique and rate two pairs of congruent instructions in English and French which were generated by drafter (see Figure 3) These texts cover most of the linguistic features of the genre of instructions (revealed from a detailed corpus analysis [1]) The evaluators were asked to focus their criticisms on issues of terminology, style, structure and grammar, and to give an overall grade on the scale: A High quality: no revision needed C Fair quality: substantial revision needed B Good quality: minor revision needed D Poor quality: needs to ....

Hartley, A. and Paris C. (1996). Two Sources of Control over the Generation of Software Instructions. In Proceedings of the 1996 Meeting of the Association for Computational Linguistics (ACL'96). Santa Cruz, California, USA.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC