A. Mili, R. Mili, and R. Mittermeir. A survey of software reuse libraries. Annals of Software Engineering, 1998. To appear.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....It does not have a forgetting mechanism incorporated to decide when to remove from user models those components that have not been reused by software developers for a long time because they might have forgot the components. 5 Related Work Many reuse repository systems (for a nice survey, see [13]) have been developed, and most of them have focused on indexing and retrieval mechanisms. Active reuse repository systems focus on a new interaction style that enables software developers to reuse components without a conscientious mode change between the reuse process and the development ....

Mili, A., Mili, R., and Mittermeir, R.T.: A survey of software reuse libraries, in Frakes, W., (ed.) Systematic software reuse. 317-347, Baltzer Science, Bussum, The Netherlands, 1998.

....up a component repository. As a response to this issue, we added the Illustrator agent to explore the decentralized approach [9, 21] to enrich the component repository by supporting the location of examples developed by peer programmers. RELATED WORK Most reusable component repository systems [10, 17] have been developed as standalone systems. CodeBroker distinguishes itself by using the information delivery mechanism to create a seamless integration between learning components and programming. A number of intelligent information systems support information delivery. Remembrance Agent [22] ....

Mili, A., Mili, R., and Mittermeir, R.T. A Survey of Software Reuse Libraries, in Systematic Software Reuse, W. Frakes, ed. Baltzer Science Publishers: Bussum, The Netherlands, 1998, 317-347.

....of the code required. These conditions can be used to search code in a data base of candidate reuse components. The pre and post condition search approach to reuse was first suggested by Katz et al. 34] Perry [45] and Rollins and Wing [46] A survey of the approach is presented by Mili [38]. Systems which automate the search for such components typically use theorem proving to find possible matches [39, 44] Unfortunately, in some cases there may be no exact match, but there may be several components which represent near matches. Fischer et al. 17] report retrieval rates of 0.49, ....

MILI, A., MILI, R., AND MITTERMEIR, R. T. A survey of software reuse libraries. Annals of Software Engineering 5 (1998), 349--414. Systematic Software Reuse.

....to the query submitted by users. The retrieval mechanism of CodeBroker is similar to that of those systems using free text indexing [12] Many sophisticated retrieval mechanisms have been proposed, such as multifaceted classification, frames and semantic networks, and associative networks [13]. Despite their sophistication and the simplicity of free text indexing, no significant difference is found in retrieval effectiveness [14] CodeBroker is unique because it is active and it strives to improve the relevance to the task and user, not to the query per se. 6 System Evaluation To ....

A. Mili, R. Mili, and R. T. Mittermeir. A survey of software reuse libraries. In W. Frakes, (ed.), Systematic Software Reuse, Annals of Software Engineering 5, pp. 317--347. Baltzer Science, Bussum, The Netherlands, 1998.

....et al. 1995; Mili et al. 1997; Schumann and Fischer, 1997; Fischer et al. 1998) It allows arbitrary specifications as search keys and retrieves all components from a library whose indexes satisfy a given match relation with respect to the key. However, in spite of all research efforts (cf. (Mili et al. 1998) for a detailed survey) it is still far away from being practicable. Notwithstanding all progress in automated deduction, the required theorem proving capabilities remain the main bottleneck. Here, we investigate a more practical approach, specification based browsing of component libraries. Its ....

....The lattice even indicates situations where additional abstractions are required to build a better index. fischer.tex; 9 08 1999; 17:16; p.2 Specification Based Browsing Of Software Component Libraries 3 2. Browsing vs. retrieval Library browsing and retrieval are closely related but following Mili et al. 1998) a clear distinction can be made. Retrieval consists of extracting components that satisfy a predefined matching criterion. Its main operation is thus the satisfaction check or matching. The criterion is usually given by an arbitrary user defined search key or query which is matched against the ....

Mili, A., R. Mili, and R. Mittermeir: 1998, `A Survey of Software Reuse Libraries'. Annals of Software Engineering 5, 349--414.

....component retrieval in general is to identify and locate potentially reusable candidates within a component library. This is a core task in software reuse: after all, components have to be found before they can be reused. A wide variety of different retrieval approaches has been investigated (cf. [13] for an overview) the approaches differ substantially in which facet of a software component (e.g. documentation, syntactic structure, execution examples) they use for retrieval and, consequently, in the mechanism to establish potential reusability. However, only deductive retrieval can exploit ....

A. Mili, R. Mili, and R. Mittermeir. "A survey of software reuse libraries", Annals of Software Engineering, 5:349-- 414, 1998.

....ambitious of these approaches is specification based retrieval [21, 22, 19, 27, 5] It allows arbitrary specifications as search keys and retrieves all components from a library whose indexes satisfy a given match relation with respect to the key. However, in spite of all research efforts (cf. [20] for a detailed survey) it is still far away from being practicable. Notwithstanding all progress in automated deduction, the required theorem proving capabilities remain the main bottleneck. Here, we investigate a more practical approach, specification based browsing of component libraries. Its ....

....formulas. An intelligent choice of such abstractions can thus speed up and improve understanding. The lattice even indicates situations where additional abstractions are required to build a better index. 2 Browsing vs. retrieval Library browsing and retrieval are closely related but following [20] a clear distinction can be made. Retrieval consists in extracting components which satisfy a predefined matching criterion. Its main operation is thus the satisfaction check or matching. The criterion is usually given by an arbitrary user defined search key or query which is matched against the ....

A. Mili, R. Mili, and R. Mittermeir. A survey of software reuse libraries. Annals of Software Engineering, 1998. To appear.

....the ILF system proved to be suitable. 2 Application Background Component retrieval is one of the technical key issues in software reuse: You must nd it before you can reuse it 1 A variety of di erent approaches has been investigated, deduction based retrieval being the most ambitious (cf. [9] for an overview. In contrast to the other approaches, it exploits exact semantic information about the components and retrieves proven matches only. Its basic idea is very simple. 1. Each component c is associated with a contract, a formal speci cation which captures the relevant behavior in ....

A. Mili, R. Mili, and R. Mittermeir. "A Survey of Software Reuse Libraries". Annals of Software Engineering, 1998. To appear.

....tactics) was used to generate the derived features. The results were evaluated to see how close it comes to implementing the classification scheme. 6.4.2. Relevance Conditions The choice of a relevance condition is fundamental in determining the significance of precision and recall measures [22]. In our experiment, we evaluated the performance of the system with respect to two relevance conditions. First, we use Satisfies match to be consistent with standard specification based retrieval experiments [21, 37] Second, we consider a relevant component to be one that matches the query ....

.... solution is to build a timeout option into the rewriting system, similar to that used by Schuman and Fischer [37] Finally, it should be noted that, in general, specification based component retrieval is susceptible to loss of recall due to the semantic gap between a component and a specification [22]. A component is associated with a specification if the component correctly implements a specification. However, there is a gap that a query may fall into: it is possible that a component may satisfy a query that its specification does not satisfy. The effects of this situation cannot be evaluated ....

A. Mili, R. Mili, and R. T. Mittermeir. A survey of software reuse libraries. Annals of Software Engineering, 5, 1998.

....most ambitious of these approaches is speci cation based retrieval [21, 22, 18, 26, 6] It allows arbitrary speci cations as search keys and retrieves all components from a library whose indexes satisfy a given match relation with respect to the key. However, in spite of all research e orts (cf. [19] for a detailed survey) it is still far away from being practicable. Notwithstanding all progress in automated deduction, the required theorem proving capabilities remain the main bottleneck. Here, we investigate a more practical approach, speci cationbased browsing of component libraries. Its ....

....formulas. An intelligent choice of such abstractions can thus speed up and improve understanding. The lattice even indicates situations where additional abstractions are required to build a better index. 2 Browsing vs. Retrieval Library browsing and retrieval are closely related but following [19] a clear distinction can be made. Retrieval consists in extracting components which satisfy a prede ned matching criterion. Its main operation is thus the satisfaction check or matching. The criterion is usually given by an arbitrary user de ned search key or query which is matched against the ....

A. Mili, R. Mili, and R. Mittermeir. "A Survey of Software Reuse Libraries ". Annals of Software Engineering, 1998. To appear.

....Put in other words, they will use a component only, if they can trust that it satisfies the expectations put into this component, i.e. it behaves as described. This brings us back to the question of how to describe reusable components. An extensive survey about how to describe reusable software [ Mili et al. 1998 ] lead us to conclude that most of the approaches discussed there are too weak to achieve both, high recall and high precision. This is not to be blamed on the methods analyzed there. As shown in [ Mittermeir et al. 1998 ] it is due to the inherent problem of describing software by ....

A. Mili, R. Mili, and R. T. Mittermeir. A survey of software reuse libraries. Annals of Software Engineering, 5:349 -- 414, 1998.

....developed in the field of software testing. The paper concludes by discussing some open problems and further extensions of this approach. 2 Issues of Uncertainty We refer to the well known correspondence between the information retrieval problem in general and software retrieval in particular [5, 6, 12]. Thus, we acknowledge the fact that a query into a repository of software components will in general yield a set of components that more or less qualify with respect to the intent of the requester, while another set of component that might also (more or less) qualify are not retrieved. This more ....

....according to preconceived categories. Thus, chances to arrive at matching abstractions for matching concepts will be improved. Nevertheless, these abstractions rely basically on human intuition. To perform them automatically has been attempted [8] But these attempts soon reach their limits [12]. 2. Signature Matching [15, 24, 25] on the contrary, is rather a formal approach, benefiting from the fact that software, even if considered as text, is a highly structured text. Basically, it provides a description on the interface level. If performed naively, one would arrive at a match ....

A. Mili, R. Mili, and R.T. Mittermeir. A Survey of Software Reuse Libraries. Annals of Software Engineering, to appear, 1998.

....affect the cost of adaptation. Our approach differs from all of the above by the fact that it requires no preliminary estimates, uses available functional information and that it produces a ranking of candidates rather than quantitative estimates. 7.3. 3 Retrieving Assets in Reuse Libraries In [35], we had compiled a survey of software storage and retrieval methods, in which we have classified methods into six families: information retrieval methods, descriptive methods, operational semantics methods, denotational semantics methods, topological methods, and structural methods. We had ....

A. Mili, R. Mili, and R. Mittermeir. A survey of software reuse libraries. Annals of Software Engineering, 1998.

....Features A software library can be characterized by a number of features, discussed below. While these attributes are not strictly orthogonal, they do give a sense of the complex issues that arise when one defines a software library. These issues are discussed in detail in Mili, Mili and Mili [1]. 1. Nature of the Asset. The most important feature of a software library is, of course, the nature of assets that are stored therein. The most typical asset is code (be it source code or executable code) but other kinds of assets are also possible: specifications, designs, test data, ....

....reuse in general, in which they discuss the storage and retrieval of software assets for the purpose of reuse. Because software libraries are not the focus of their surveys, they are treated at a general level, and account for the state of the art at their time of publication. Mili, Mili and Mili [1] present a survey on software reuse libraries, in which they survey the field, identify software library characteristics, discuss the different techniques for library organization, and review the literature on software library access methods. Asset retrieval and browsing are method for accessing ....

H. Mili, F. Mili, and A. Mili. A survey of software reuse libraries. Annals of Software Engineering, 5:349--414, 1998.

No context found.

A. Mili, R. Mili, and R. Mittermeir. A survey of software reuse libraries. Annals of Software Engineering, 1998. To appear.

No context found.

A. Mili, R. Mili, and R. Mittermeir. A survey of software reuse libraries. Annals of Software Engineering, Vol. 5:pages 349--414, 1998.

No context found.

:445--460. Mili, A.; Mili, R.; and Mittermeir, R. 1998. A survey of software reuse libraries. Annals of Software Engineering

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