FRAKES,W.B.AND NEJMEH, B. A. 1987. Software reuse through information retrieval. In The 20th Annual HICSS. Vol. 2, Software, B. D. Shriver, Ed. Western Periodicals Company, 530--535.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....service retrieval approach is minimizing the manual effort involved in listing new services with the search engine. Ideally services can be classified automatically. Previous efforts for automatic service classification have used similarity metrics based mainly on word frequency statistics [35] [36] 37] We plan to augment such approaches using the taxonomic reasoning and graph theoretic similarity measures made possible by our approach to service modeling. User interface. PQL, like most query languages, is fairly verbose and requires that users be familiar with its syntax. An ....

Frakes, W.b. and B.a. Nejmeh, Software reuse through information retrieval. Proceedings of the Twentieth Hawaii International Conference on System Sciences, 1987. 2: p. 6-9.

....service retrieval approach is minimizing the manual effort involved in listing new services with the search engine. Ideally services can be classified automatically. Previous efforts for automatic service classification have used similarity metrics based mainly on word frequency statistics [35] [36] 37] We plan to augment such approaches using the taxonomic reasoning and graph theoretic similarity measures made possible by our approach to service modeling. User interface. PQL, like most query languages, is fairly verbose and requires that users be familiar with its syntax. An ....

Frakes, W.b. and B.a. Nejmeh, Software reuse through information retrieval. Proceedings of the Twentieth Hawaii International Conference on System Sciences, 1987. 2: p. 6-9.

....using IR methods to support software engineering tasks and recovering source code to documentation links. The research that has been conducted on the specific use of applying information retrieval methods to source code and associated documentation typically relates to indexing reusable components [12, 13, 16, 17]. Notable is the work of Maarek [16, 17] on the use of an IR approach for automatically constructing software libraries. The success of this work along with the inefficiencies and high costs of constructing the knowledge base associated with natural language parsing approaches to this problem [10] ....

Frakes, W., "Software Reuse Through Information Retrieval", in Proceedings 20th Annual HICSS, Kona, HI, Jan. 1987, pp. 530-535.

....Clustering of source code based on semantic and structural information is very useful in the maintenance and evolution of legacy software systems. For instance, the clustering can be used to assist in the re modularization [30, 34, 35, 39] of systems and the identification of abstract data types [8, 17]. If a software system were to be reengineered into an object oriented language from a structured one, then this type of clustering would prove to be very useful. The objective is to reduce the amount of source code an engineer needs to view, at any one time, and give them clues about possible ....

Frakes, W., "Software Reuse Through Information Retrieval", in Proc. of 20th Annual HICSS, Kona, HI, Jan. 1987, pp. 530-535.

....critical of the above issues. P. Freeman [25] stresses the importance of locating items: Our continued work in this area, however, has convinced us that we had assumed away one of the more critical aspects of reusability locating an appropriate set of information to reuse . W. Frakes et al. [23] make the statement: there is strong support for the belief that to effectively promote software reuse we must develop tools to aid in the process of locating software components that are candidates for reuse . Libraries should and will increase in size. Finding a relevant component is ....

W. B. Frakes, B. A. Nejmeh. Software Reuse Through Information Retrieval. Pro- ceedings of the 12th Annual Hawaii International Conference on system Sciences pp 530-535. (1987)

....profile. A profile is an abbreviated description of the original document that is easier to manipulate. The research that has been conducted on the specific use of applying information retrieval methods to source code and associated documentation typically relates to indexing reusable components [15, 16, 18, 28, 29, 32, 35]. Notable is the work of Maarek [28, 29] on the use of an IR approach for automatically constructing software libraries. The success of this work along with the inefficiencies and high costs of constructing the knowledge base associated with natural language parsing approaches to this problem [12, ....

.... Most of them use formal features (i.e. structural information) and two of them ( 28] and [37] use semantic information (referred to as non formal descriptive features) The research that has been conducted on the specific use of applying information retrieval methods to source code includes [15, 16, 18, 2830, 32, 35]. 7. Conclusions The experiments with the PROCSSI system show that the semantic similarity of source code documents provides valuable information that can be used in the tasks of software maintenance and evolution. It also shows that concepts from the problem domain are often spread over ....

Frakes, W., "Software Reuse Through Information Retrieval", in Proceedings of 20th Annual HICSS, Kona, HI, Jan. 1987, pp. 530-535.

....with the code at all. This is particularly an issue when the code is poorly commented since the query words would no longer likely match words in the natural language code comments. In the second category, we nd tools that are based on natural language documentation associated with the code [6, 8, 15]. It is worthwhile noting that such tools are designed for retrieval of reusable components whereas our tool is a browsing and search tool for applications. Moreover, as discussed earlier, natural language documentation may not be available. Our use of the application GUI takes advantage of ....

W. B. Frakes and B. A. Nejmeh. Software reuse through information retrieval. In 20th Hawaii International Conference on System Sciences, pages 530535. IEEE, 1987.

....model and to implement querying facilities that retrieve them for potential reuse based on a user request. More in general, means to trace source code to free text documents are useful whenever it is required that bottom up structured analysis be combined with top down or modeldriven analysis [8, 3, 18]. We propose an approach to establish and maintain traceability links between the source code and free text documents. A premise of our work is that programmers use meaningful names for program s items, such as functions, variables, types, classes, and methods. We believe that the ....

....based on free text analysis is RSL [8] system which extracts free text single term indices from comments in source code files looking for reuse keywords like author , date created , etc. REUSE [3] is an information retrieval system which stores software objects as textual documents. CATALOG [18] is an information retrieval system to store and retrieve C components which are individually characterized by a set of single term indices which are automatically extracted from natural language headers of C programs. 6. Conclusions We have presented an approach to recover traceability links ....

W. B. Frakes and B. A. Nejmeh. Software reuse through information retrieval. In Proceedings of 20-th Ann. HICSS, Kona (HI), pages 530--535, January 1987.

....to more specific processes such as Analyze creditworthiness using scoring and Analyze creditworthiness using PE ratio, etc. This process representation has equal or greater formal expressiveness than other full fledged process modeling languages (e.g. IDEF [24] PIF [25] PSL [26] or CIMOSA [27]) as well as greater expressiveness than the frame based languages used in previous service retrieval efforts, by virtue of adding such important concepts as full typing, resource dependencies, ports, task decompositions, and exceptions. The growing Handbook database currently includes over 5000 ....

Frakes, W.b. and B.a. Nejmeh, Software reuse through information retrieval. Proceedings of the Twentieth Hawaii International Conference on System Sciences, 1987. 2: p. 6-9.

....system. Traditional approaches to support software retrieval can basically fall into two categories, automatic indexing and knowledge based. The automatic indexing approach takes the natural language documentation of the component and automatically extracts a set of indices to characterize it. [12], 14] and [20] describe systems based on this approach. With these techniques, retrieval is in general efficient, but help with selection and correct usage must be found outside the retrieval system. On the other hand, in the knowledge based approach, the characterization of the component is ....

Frakes, W. B. & Nejmeh, B. A., 1987. "Software Reuse through Information Retrieval ". Procs. of the 20th Annual HICSS, Kona, HI.

....Thus all kinds of software objects, including user menus and system thesauri, are stored as textual documents. These two systems use some kind of IR related technique. However the only system, to our knowledge, that applies a pure IR free text approach is the system proposed by Frakes and Nejmeh [Frakes and Nejmeh, 1987]. They use the catalog information retrieval system for storing and retrieving C software components. Each component is characterized by a set of single term indices that are automatically extracted from the natural language headers of C programs. Therefore, the construction of the C components ....

Frakes, W. and Nejmeh, B. (1987). Software reuse through information retrieval. In Proceedings of the 20th Annual HICSS, pages 530--535, Kona, HI.

....of the software components. It does not use any semantic knowledge, nor intend to understand the documentation. The goal of this approach is to characterize each component by a set of indices that are automatically extracted from its natural language documentation. The system proposed in (Frakes Nejmeh 1987) uses an existing IR system, CATALOG, for storing and retrieving C software components. Each component is characterized by a set of single term indices that are automatically extracted from the natural language headers of C programs. The GURU system (Maarek et al. 1991) classifies the software ....

Frakes, W. B. & Nejmeh, B. A. (1987). Software Reuse through Information Retrieval. Procs. of the 20th Annual HICSS, Kona, HI.

....topic in research on code reuse has been component based software engineering. It is Error No text of specified style in document. 2 seen as an enabling technology for successfully developing more complex software by more code reuse, leading to an increase in development productivity [Ber97] [FrN87]. Assembling systems from a choice of software components is an obvious example of code reuse. We have introduced the object oriented WebComposition model [GWG97] and the WebComposition Markup Language (WCML) GBG98] to allow a complete mapping of design concepts to implementation. WCML ....

....WCML facilitates Web application development by means of composing Web applications from components. Examining the requirements for successful code reuse it became clear that adequate libraries or repositories as well as tools for locating and retrieving code from them were required [MBK91] [FrN87]. To fully exploit the advantages of code reuse when using the WebComposition system a repository for WCML components is being built. In this paper we will give an overview of the WebComposition approach and introduce our research and development concerning the repository for WCML components. ....

[Article contains additional citation context not shown here]

W.B. Frakes, B.A. Nejmeh (1987). Software Reuse through Information Retrieval, In: Proceedings of the 20 th Annual Hawaii International Conference On System Sciences, 1987.

....given task. Finding that component among all the others has been identified as a major problem of compositional reuse. It has been found that a repository for storing components and aiding their location and retrieval is required to facilitate the reuse of components enough to make it attractive [6, 16]. Such a repository should utilize representations of the components that differ from the information extractable from the components themselves to be able to offer advanced methods for location and retrieval. To fully exploit the advantages of code reuse when using the WebComposition system we ....

....CBSE is said to allow the construction of more complex software at lower costs. It is supposed to lead to easier maintenance and evolution (i.e. a higher flexibility of a software product throughout its entire life cycle) as well as an overall increase of quality if performed systematically [2, 6, 11]. Applying Component Based Software Engineering to the Web In analogy to CBSE we introduce the term Component Based Web Engineering to denote the construction of Web Applications according to a disciplined process involving systematic compositional reuse. Definition Component Based Web ....

[Article contains additional citation context not shown here]

W.B. Frakes, B.A. Nejmeh (1987). Software Reuse through Information Retrieval, In: Proceedings of the 20 th Annual Hawaii International Conference On System Sciences, 1987.

....specification level interoperability, which is a form of module signature match, since their prototype specifications use only signature information. Less closely related work, but relevant to our context of software library retrieval, divides into two categories: text based information retrieval [7, 2, 23, 16] and AI based semantic net classifications [21, 6] The advantage to these approaches is that many efficient tools are available to do the search and match in these structures. The disadvantage is that a component s behavior is described informally. 8 Summary This paper makes three specific ....

Frakes, W. B., and Nejmeh, B. A. Software reuse through information retrieval. In The 20 th Annual HICSS, Vol. 2: Software (1987), B. D. Shriver, Ed., Western Periodicals Co., pp. 530--535.

....Most work thus far has focused on signature based interoperability, and how to convert types in a heterogeneous environment [Kon93, YS94, Tha94] Less closely related work, but relevant to our context of software library retrieval, divides into three categories. Text based information retrieval [FN87, AS87, PD89, MBK91] and AI based semantic net classifications [OHPDB92, FHR91] have the advantage that many efficient tools are available to do the search and match in these structures. The disadvantage is that a component s behavior is described informally. A third class of retrievals [PP94, ....

W. B. Frakes and B. A. Nejmeh. Software reuse through information retrieval. In Bruce D. Shriver, editor, The 20 th Annual HICSS, Vol. 2: Software, pages 530--535. Western Periodicals Co., 1987.

....on traditional document retrieval systems, where keywords can be searched for within text. Comments within components and any associated documentation would provide the scope for a keyword search. Examples of retrieval systems based on keywords have been produced by Frakes and Nejmeh of AT T [Frakes86] and at Intermetrics in the Reusable Software Library (RSL) Burton87] Both these retrieval systems exhibit a number of deficiencies. Poor descriptions of components will result in poor retrievals, either because a component cannot be located when expected, or worse, it is located when it ....

Frakes, W.B., Nejmeh, B.A. "Software Reuse Through Information Retrieval", SIGIR Forum, pp.30-36 (1986).

....based on this idea, such as RSL[Burton87] which 1 UNIX is a registered trademark of Unix Systems Laboratories. 2 X Window System is a trademark of the Massachusetts Institute of Technology. 2 catalogues the software by attributes (e.g. keywords, authors . extracted from comments. CATALOG[Frakes87] recognizes indexes contained in C programs headers, coded in natural language; GURU[Maarek91] analyzes phrases contained in the software documentation and the source codes. Moreover, Prieto Diaz and Freeman [Prieto Diaz89] describe a system based on faceted schema and analyze several program ....

W.B. Frakes and B.A. Nejmeh, Software reuse through information retrieval, Proc. 20th Ann HICSS, Kona, Jan, 1987, pp. 530-535.

....Searching a software library for a particular item (source code or design documentation) is a central activity in all stages of the software development life cycle. In software reuse, for example, the locating of relevant software artifacts has long been recognized as an essential activity [3, 19, 18]. At present, it is a slow knowledge intensive process that is by no means guaranteed to succeed, and almost all aspects of software development are hampered by the cost and low success rate of this search. As libraries grow so does the problem. A person s search through a software library is ....

W. B. Frakes, B. A. Nejmeh. Software Reuse Through Information Retrieval. Proceedings of the 12th Annual Hawaii International Conference on system Sciences pp 530--535. (1987)

....for all components within the software library that contain certain words specified by the user. The text searched can either be source code, or a short text description of a component stored in the on line library. An example of this approach is that used by the CATALOG system from AT T [59][53]. The full text retrieval approach avoids the need for manual indexing, which is costly and can be inconsistent. However, as the library becomes larger, it becomes difficult to search and retrieve all the relevant resources, and to differentiate those from non relevant resources [63] Special ....

Frakes, W., and Nejmeh, B., "Software Reuse Through Information Retrieval," Proceedings of the Twentieth Annual Hawaii International Conference on System Science,eds. B.D.Shriver and R.H.Sprague, Jr., Kailua-Kona, Hawaii, IEEE Computer Society, Los Alamitos, CA, 1987, pp. 530-535.

....into tools for software reuse, and we discuss some of them in what follows. Perhaps the most well known approaches are those concerned with finding a suitable component (such as a function or class) in a library that fits a particular need. Such work includes tools that use free text indexing [2], facets [5] signature matching [7] and formal specifications [1] While our approach is not query based, browsing reuse boundary intersections for example programs would show important classes and relationships between them in the library even if the developer does not know exactly what ....

W. B. Frakes and B. A. Nejmeh. Software Reuse through Information Retrieval. In 20th Hawaii International Conference on System Sciences, pages 530--535. IEEE, 1987.

....Bottom up approaches require that the user select and assemble architecturally compatible sets of components from a software library. For example, component retrieval tools are concerned with finding components that fit a particular need. Such work includes tools that use free text indexing [11], facets [21] and specification matching [18] Although these tools may help a developer find individual components of interest, they do not show how these components can be used in combination. There are also tools that help a developer examine a software library in terms of architecture, ....

W. B. Frakes and B. A. Nejmeh. Software reuse through information retrieval. In 20th Hawaii International Conference on System Sciences, pages 530-- 535. IEEE, 1987.

....that searching a software library for a particular item (source code or design documentation) is a central activity in all stages of the software development life cycle. In software reuse, for example, the locating of relevant software artifacts has long been recognized as an essential activity [5, 21, 20, 18]. At present, it is a slow knowledge intensive process that is by no means guaranteed to succeed, and almost all aspects of software development are hampered by the cost and low success rate of this search. As libraries grow so does the problem. Research aimed at improving the speed and 1 ....

W. B. Frakes and B. A. Nejmeh.(1987) Software Reuse Through Information Retrieval. Proceedings of the Twelfth Annual Hawaii International Conference on System Sciences pp 530--535.

....Repositories 11 23 99 1 Henninger 1. Introduction As libraries of reusable software components continue to grow, the issue of retrieving components from software libraries has captured the attention of the software reuse community [Burton et al. 1987; Devanbu et al. 1991; Frakes, Gandel 1990; Frakes, Nejmeh 1987; Frakes, Pole 1994; Maarek et al. 1991; Prieto Daz, Freeman 1987; Sommerville, Wood 1986] Especially in component based reuse approaches, where developers compose software parts into an application, libraries of components are necessary to achieve software reuse. Component based software reuse ....

....but when many examples are available, finding and choosing appropriate ones becomes troublesome. Retrieval techniques as diverse as enumerated classification [Booch 1987] facets [PrietoD az 1985; Prieto Daz, Freeman 1987] frame based classification [Ostertag et al. 1992] free text indexing [Frakes, Nejmeh 1987], relational databases [Burton et al. 1987] and formal specifications [Chen et al. 1993] have been employed to address the problem of finding relevant components. But issues involving how effective repositories are built, populated, and evolved to meet the changing needs of development ....

Frakes, W. B., Nejmeh, B. A. "Software Reuse Through Information Retrieval," The 20th Hawaii International Conference on System Sciences, Hawaii, IEEE Computer Society, Los Alamitos, CA, pp. 530-535.

....and Our Suggestion One of the influential systems applying techniques of information retrieval (IR) in retrieving software components is the one developed by Prieto Diaz (cf. 13] which classifies software through detailed analysis. This work has attracted great attention. Systems described in [6], 15] 10] utilize the same principle. These approaches present advantages in terms of cost, portability, and scalability. Information about the software can be obtained from the documentation of the software (such as in [10] or from the categories that a software belongs to (such as in [15] ....

Frakes, W. B. & Nejmeh, B. A.: Software Reuse through Information Retrieval. In: Proceedings of the 20th Annual HICSS, Kona, HI, pp. 530-535, 1987.

....into tools for software reuse, and we discuss some of them in what follows. Perhaps the most well known approaches are those concerned with finding a suitable component (such as a function or class) in a library that fits a particular need. Such work includes tools that use free text indexing [3], facets [8] signature matching [11] and formal specifications [2] While our approach is not query based, browsing reuse boundary intersections for example programs would show important classes and relationships between them in the library even if the developer does not know exactly what ....

W. B. Frakes and B. A. Nejmeh. Software Reuse through Information Retrieval. In 20th Hawaii International Conference on System Sciences, pages 530--535. IEEE, 1987.

....for classification for the description of complex domains. Since components usually have orthogonal aspects describing their properties the classification of components according to more than one aspect in parallel seems appropriate. There exist several recent approaches in this direction ([10], 13] 14] 17] Most of them rely on static classification schemes and or consider only restricted kinds of component properties. Our approach supports arbitrary kinds of components, as well as changing classification schemes. We think this is important, since classification schemes tend to ....

....relationship. The candidates revealed by a query to the database are then processed formally according to their type signature as proposed in [6] 8] In automatic classification information retrieval methods are used to compute descriptors automatically from textual descriptions of components ([10]) This frees users from maintaining the classification scheme. But such methods do only work, if there exist textual descriptions to apply this method to. Since there is no explicit classification scheme, there is neither a compact information about the content of the library, nor a useful basis ....

W.B. Frakes, B.A. Nejmeh, "Software Reuse Through Information Retrieval," Proceedings 32nd COMPCON '87, IEEE CS Press, Los Alamitos, CA, 1987, 380--384.

....applications, such as resource allocator, customer queue, etc. There is thus a distintion between what a software component is as such and what it reflects in the outside world. Various techniques have been suggested for software component retrieval, such as uncontrolled indexing (e.g. [14, 9]) enumerated classification (e.g. 1, 4] faceted classification (e.g. 17] thesaurus [3] and knowledge representation techniques (e.g. 18, 8] Some approaches can be hybrids of several techniques. An example of this is [15] which uses faceted classification, but at the same time facets ....

W. B. Frakes and B. A. Nehmeh. Software reuse through information retrieval. SIGIR FORUM, 21(1--2):30--36, 1987.

....similarity matching. The name matching method matches those components that have the same standardized name in each library. The similarity matching method uses more conventional information retrieval techniques and is similar to that used in component retrieval tools based on free text indexing [5, 8, 11]. In this paper we shall be concerned with components that are classes or functions. To simplify the exposition, we will just say component whenever the discussion applies to both classes and functions. To avoid ambiguity, keep in mind that we will only match classes with classes and functions ....

....exactly what we do in Section 4. But name matching alone is not enough. It is still possible that some important concept is represented by components with completely different names in the libraries. For this reason, we consider well known component retrieval techniques such as free text indexing [5], facets [14] and formal specifications [2] However, since we are interested in automated techniques that do not require domain analysis or formal specifications in the code, we exclude facets and formal specifications from our discussion. The free text indexing method simply uses the text in ....

W. B. Frakes and B. A. Nejmeh. Software Reuse through Information Retrieval. In 20th Hawaii International Conference on System Sciences, pages 530--535. IEEE, 1987.

....a tool we have built for assessing C and Java libraries. 1 Introduction Selecting from among a set of candidates is an important step in software reuse [7] There are a number of approaches for selecting from among relatively small reuse candidates such as individual methods or classes [3, 4, 9, 12]. These approaches are generally query based, which usually works well in part because there are often a reasonably large number of candidates. But selecting among larger candidates such as selecting a thread library or a user interface framework is difficult and potentially very expensive ....

....With existing tools, the user must manually integrate the knowledge learned about each library. In contrast, there are numerous reuse tools that help a developer select a fine grained component (such as a function or class) in a particular library. These include tools that use free text indexing [4], facets [9] signature matching [12] and formal specifications [3] As noted above, such techniques do not carry over to library selection because: a) the libraries under consideration usually have similar functional but different non functional properties, while the components in a particular ....

W. B. Frakes and B. A. Nejmeh. Software Reuse through Information Retrieval. In 20th Hawaii International Conference on System Sciences, pages 530--535. IEEE, 1987.

....and similarity matching. The name matching method matches those components that have the same standardized name in each library. The similarity matching method uses more conventional information retrieval techniques similar those used in component retrieval tools based on free text indexing [4, 7, 10]. In this paper we shall be concerned with components that are classes or functions. We use the term library to mean any collection of reusable components such as a toolkit or framework. To simplify the exposition, we will just say component whenever the discussion applies to both classes and ....

....sets of libraries. Another approach is to exploit formal specifications associated with the code to identify similar behaviors [12] since most libraries do not have formal specifications associated with them, the approach is either limited or costly. A third approach is to use free text indexing [4], which uses the text in the libraries (and not just the component names) for indexing using standard information retrieval techniques [3] No manual domain analysis is required, nor must the libraries be annotated with additional information. Researchers have observed that this method works well ....

W. B. Frakes and B. A. Nejmeh. Software reuse through information retrieval. In 20th Hawaii International Conference on System Sciences, pages 530--535. IEEE, 1987.

....its retrieval effectiveness as developers use it to find reusable software artifacts. 1. Introduction As libraries of reusable software components continue to grow, the issue of retrieving components from software component repositories has captured the attention of the software reuse community [5, 8, 12, 13, 21, 25]. Methods using these repositories face an inherent dilemma: in order for the approach to be useful, the repository must contain enough components to support developers, but when many examples are available, finding and choosing an appropriate one becomes troublesome. Retrieval techniques as ....

....the repository must contain enough components to support developers, but when many examples are available, finding and choosing an appropriate one becomes troublesome. Retrieval techniques as diverse as enumerated classification [1] facets [25] frame based classification [23] free text indexing [13], and relational databases [5] have been employed to address the problem of finding relevant components. But issues involving how effective repositories are built, populated, and evolved to meet the changing needs of developers have received considerably less attention. Most retrieval algorithms ....

W. B. Frakes, B. A. Nejmeh, "Software Reuse Through Information Retrieval," The 20th Hawaii International Conference on System Sciences, Hawaii, 1987, pp. 530-535.

....Bottom up approaches require that the user select and assemble architecturally compatible sets of components from a software library. For example, component retrieval tools are concerned with finding components that fit a particular need. Such work includes tools that use free text indexing [3], facets [11] and specification matching [8] Although these tools may help a developer find individual components of interest, they do not show how these components can be used in combination. There are also tools that help a developer examine a software library in terms of architecture, style, ....

W. B. Frakes and B. A. Nejmeh. Software reuse through information retrieval. In 20th Hawaii International Conference on System Sciences, pages 530--535. IEEE, 1987.

No context found.

FRAKES,W.B.AND NEJMEH, B. A. 1987. Software reuse through information retrieval. In The 20th Annual HICSS. Vol. 2, Software, B. D. Shriver, Ed. Western Periodicals Company, 530--535.

No context found.

W. B. Frakes and B. A. Nejmeh. Software Reuse through Information Retrieval. SIGIR Forum, 21(1-2):30--36, 1987.

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