Automated Software Engineering:
An International Journal

Special Issue on

Inductive Programming

Guest Editors: Pierre Flener and Derek Partridge


The first step in idealised software engineering is to abstract a precise specification, which is then taken as the foundation for all subsequent development, such as coding and testing. However, complex specifications typically contain errors and approximations that lead to errors in the eventual software, errors that are not easily detected or eliminated before the software is subjected to operational testing.

Many programming problems are manifest as sets of data values, namely inputs and corresponding outputs. Inductive programming techniques work from such data instances to the implementation without going through a specification. They thus offer the software engineer a means to avoid or rectify specification-induced system errors and perhaps even to circumvent the need for a complete a priori specification, i.e., some system modules may be inductively generated from data where accurate specification proves difficult. Induction-based processing of problem data may be used to check and correct features of a specification. Inductive software development will be particularly germane, if not essential, for complex data-defined problems, or problem modules. These will arise in a data-rich domain and address very complex aspects of the world, such as the human body, complicated manufacturing processes, and complex dynamical situations.

Inductive programming is not, however, trouble-free. There are difficult issues of `understanding' the automatic induction procedures so that implementation performance can be characterised. There are issues of data pre-processing to facilitate optimal application of a given inductive technique and to obtain an implementation with certain desired characteristics.


The intent of this special issue is to bring together developments in inductive programming that have a direct bearing on software development. Papers are sought on novel uses of such inductive techniques, whether automatic or interactive, for practical software development. It is anticipated that relevant work is to be found in a variety of different areas, including rule induction, neural computing, data mining, machine learning, and inductive logic programming.

We solicit papers about, but not limited to, the following topics:

Submission Guidelines

Accepted papers must be prepared using the ASE author instructions (available from the ASE homepage).

Although there is no restriction on length, we would prefer shorter papers to longer ones, for the sake of greater diversity and more thorough reviewing. Authors are therefore encouraged to be as concise as possible, and to limit their papers to 30 pages.

Electronic submissions are encouraged, and may be sent as one email (to both guest editors), MIME attachments being allowed. The message should contain (i) the abstract and author names and addresses in ASCII, and (ii) the whole paper in Postscript or PDF. A Postscript version must be interpretable by Ghostscript, and must use standard fonts, or include the necessary fonts. Authors who cannot meet these requirements should submit 5 hard copies by post instead (to either of the guest editors).

All submitted papers will be refereed according to the usual ASE refereeing process.

To aid planning and organisation, we would appreciate an email or a letter of intent to submit a paper (including author information, a tentative title and abstract, and an estimated number of pages) as early as possible.

Important Dates

Submission: 30 November 1999
Final notification:15 September 2000
Publication: early 2001

Guest Editors

Pierre Flener, Uppsala University, Sweden
Derek Partridge, University of Exeter, UK