Patent Application
20090182758
The present application is related to a commonly assigned U.S. application entitled “Method for Automatically Computing Proficiency of Programming Skills,” identified by attorney docket number IN920070074US1, and filed on even date herewith, the disclosure of which is incorporated by reference herein in its entirety.
The present invention generally relates to information technology, and, more particularly, to proficiency assessment.
Challenges exist is the area of assessing proficiency of programming skills. Existing approaches assess proficiency manually by human assessors. Existing approaches also include a high operation cost, especially when a large number of invidivuals are being assessed on an ongoing basis (because people's skills evolve). However, there also exists a high cost for not performing proficiency assessments. Neglecting such assessments can lead to improper or detrimental matching of skills to project requirements.
Principles of the present invention provide techniques for automatically computing proficiency of programming skills from programmer artifacts.
An exemplary method (which may be computer-implemented) for automatically computing a programmer proficiency rating for one or more programmers, according to one aspect of the invention, can include steps of obtaining one or more programmer artifacts for each programmer to be assessed, obtaining one or more programmer artifacts and one or more human proficiency ratings for a separate set of one or more programmers, training a first module to learn a rating model from the one or more programmer artifacts and one or more human proficiency ratings for the separate set of one or more programmers, and using a second module to apply the rating model to the one or more programmer artifacts for each programmer to be assessed to automatically generate the programmer proficiency rating for each programmer.
In an embodiment of the invention, an exemplary method for generating a database of one or more programmer proficiency ratings includes the following steps. One or more programmer artifacts for each programmer are obtained. Data analysis is performed on the one or more programmer artifacts to compute one or more program quality features. The one or more program quality features and one or more classification techniques are used to compute a programmer proficiency raring for one or more programmers. Also, the programmer proficiency rating is stored in a searchable database.
At least one embodiment of the invention can be implemented in the form of a computer product including a computer usable medium with computer usable program code for performing the method steps indicated. Furthermore, at least one embodiment of the invention can be implemented in the form of an apparatus including a memory and at least one processor that is coupled to the memory and operative to perform exemplary method steps.
These and other objects, features and advantages of the present invention will become apparent from the following detailed description of illustrative embodiments thereof, which is to be read in connection with the accompanying drawings.
Principles of the present invention include assessing technical skill levels of information technology (IT) programmers. One or more embodiments of the invention include using automatically computed program quality features, as well as using classifiers to learn programmer proficiency from training data. Additionally, principles of the invention include computing the proficiency of a programmer from the programmer artifacts that are created in the normal course of software development.
As described herein, principles of the invention include automatically assessing proficiency of programming skills of individuals using statistical learning techniques. The techniques detailed herein greatly reduce the need for human (that is, manual) assessment of programming skills of individuals, and lead to better matching of individuals to project requirements (for example, in a software group or in a services group).
One or more embodiments of the present invention improve the uniformity of assessment across an organization, minimize human effort required for ranking practitioners, and also can be implemented as an application to various organizations.
As illustrated in
The step of applying a classifier trainer can be iterated, for example, until the rating model converges for given classifier trainer. Also, the output of a PRTM is a rating model.
As illustrated in
One difference between
The classifier 210, in contrast, applies the previously learnt rating model 114 (or 212) to automatically generate programmer proficiency ratings 214 from program quality features 108 (or 208), which are in turn generated by the data analysis module 106 (or 206) that analyzes programmer artifacts 102 (or 202).
During the operation of the PRTM operation, the PRTM infers a relationship between the program quality features and proficiency ratings by humans for a subset of the programmers. This relationship is encoded within the rating model. The rating model is the output of the PRTM, and is used by the PRM.
Operating the PRM includes outputting a proficiency rating for a programmer using the programmer artifacts. For example, an organization has 10,000 programmers. A small subset of 1,000 programmers (10%) are rated by humans. The PRTM would use programmer artifacts and humans ratings of these 1,000 programmers to output the rating model. The PRM would use this rating model to compute the programmer proficiency ratings for all 10,000 programmers, including the 9,000 that were unassessed by humans.
With a properly designed PRTM and PRM, the PRM outputs a proficiency rating close to what a human assessor would have typically assigned (and as part of the classifier training, this is checked for the 1,000 available human assessments), while ironing out the variations between human assessors.
The PRTM is used to output the rating model, and thereafter used periodically to update or tune the rating model as additional or fresh assessments by humans are made available.
As described herein, one or more embodiments of the present invention include programmer artifact(s), classifier trainer(s), classifier(s), rating model(s), programmer proficiency rating(s), and programmer proficiency rating(s) by humans. Programmer artifacts may include, for example, design documents, programs (that is, code), etc. written by a developer (for example, in the past few months or years) that may also be filtered by language and/or platform. A classifier trainer may include training modules for classifiers such as, for example, a support vector machine (SVM), linear classifiers, maximum entropy, neural networks, etc.
A classifier may include run-time classification modules for classifiers such as, for example, SVM, linear classifiers, maximum entropy, neural networks, etc. A rating model may include a trained model output by a classifier trainer (for example, for SVM, linear classifiers, etc.) that is used by a corresponding classifier to obtain programmer proficiency ratings. Programmer proficiency rating includes a rating of the programming skill of a programmer (for example, on a scale of 1-5 (5 being a skilled programmer, and 1 being a novice programmer). Also, programmer proficiency rating(s) by humans include a programmer proficiency rating (as described above) assessed by a human.
One or more embodiments of the present invention may also include data analysis and program quality features. Data analysis may include, for example, a module that computes program quality features used by classifier trainers and classifiers using programmer artifacts.
Program quality features include features (that is, statistics or any computed quantity) that convey useful information about the quality of programs. Such features may include, for example, average number of classes used, number of global variables used, number of static variables used, number of lines of code per method, number of side effects of methods, number of private and public instance variables, interfaces used, inherited classes used, inner classes used, etc. Additional features may include, for example, defect rates (for example, standard measures such as defects per kilo-line of code or defects per function point).
Step 306 includes training a first module (for example, a PRTM) to learn a rating model from the one or more programmer artifacts and one or more human proficiency ratings for the separate set of one or more programmers. Training the first module can include performing a data analysis on the one or more programmer artifacts to compute one or more program quality features, and using a classifier trainer to learn a rating model from the program quality features and proficiency ratings by human assessors for the separate set of programmers. Data analysis can be performed automatically by using computer programs that parse the code to identify various elements in the source code, followed by numeric computations to compute the quality features. In an illustrative embodiment of the invention, a rule-based approach may be used to identify various elements in the source code.
Also, a classifier trainer may be trained, for example, to mimick human assessors using proficiency ratings computed by humans for a subset of the one or more programmers. The classifier trainer (for example, a program) will learn to rate the proficiency of programmers from a set of previous examples.
Program quality features may include, for example, average number of classes used, average number of lines of code per method, average number of global variables used, average number of static variables used, average number of interfaces used, average number of inherited classes used, average defect rates, average number of side effects of methods, average number of private and public instance variables, average number of inner classes used and productivity measures.
Step 308 includes using a second module (for example, a PRM) to apply the (learnt) rating model to the programmer artifacts for each programmer to be assessed to automatically generate the programmer proficiency rating for each programmer. The programmer proficiency rating may include, for example, a rating of a programming skill of a programmer. Also, the techniques depicted in
A variety of techniques, utilizing dedicated hardware, general purpose processors, software, or a combination of the foregoing may be employed to implement the present invention. At least one embodiment of the invention can be implemented in the form of a computer product including a computer usable medium with computer usable program code for performing the method steps indicated. Furthermore, at least one embodiment of the invention can be implemented in the form of an apparatus including a memory and at least one processor that is coupled to the memory and operative to perform exemplary method steps.
At present, it is believed that the preferred implementation will make substantial use of software running on a general-purpose computer or workstation. With reference to
Accordingly, computer software including instructions or code for performing the methodologies of the invention, as described herein, may be stored in one or more of the associated memory devices (for example, ROM, fixed or removable memory) and, when ready to be utilized, loaded in part or in whole (for example, into RAM) and executed by a CPU. Such software could include, but is not limited to, firmware, resident software, microcode, and the like.
Furthermore, the invention can take the form of a computer program product accessible from a computer-usable or computer-readable medium (for example, media 518) providing program code for use by or in connection with a computer or any instruction execution system. For the purposes of this description, a computer usable or computer readable medium can be any apparatus for use by or in connection with the instruction execution system, apparatus, or device.
The medium can be an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system (or apparatus or device) or a propagation medium. Examples of a computer-readable medium include a semiconductor or solid-state memory (for example, memory 504), magnetic tape, a removable computer diskette (for example, media 518), a random access memory (RAM), a read-only memory (ROM), a rigid magnetic disk and an optical disk. Current examples of optical disks include compact disk-read only memory (CD-ROM), compact disk-read and/or write (CD-R/W) and DVD.
A system, preferably a data processing system suitable for storing and/or executing program code will include at least one processor 502 coupled directly or indirectly to memory elements 504 through a system bus 510. The memory elements can include local memory employed during actual execution of the program code, bulk storage, and cache memories which provide temporary storage of at least some program code in order to reduce the number of times code must be retrieved from bulk storage during execution.
Input and/or output or I/O devices (including but not limited to keyboards 508, displays 506, pointing devices, and the like) can be coupled to the system either directly (such as via bus 510) or through intervening I/O controllers (omitted for clarity).
Network adapters such as network interlace 514 may also be coupled to the system to enable the data processing system to become coupled to other data processing systems or remote printers or storage devices through intervening private or public networks. Modems, cable modem and Ethernet cards are just a few of the currently available types of network adapters.
In any case, it should be understood that the components illustrated herein may be implemented in various forms of hardware, software, or combinations thereof, for example, application specific integrated circuit(s) (ASICS), functional circuitry, one or more appropriately programmed general purpose digital computers with associated memory, and the like. Given the teachings of the invention provided herein, one of ordinary skill in the related art will be able to contemplate other implementations of the components of the invention.
At least one embodiment of the invention may provide one or more beneficial effects, such as, for example, improving the uniformity of assessment across and organization and minimizing human effort required for ranking practitioners.
Although illustrative embodiments of the present invention have been described herein with reference to the accompanying drawings, it is to be understood that the invention is not limited to those precise embodiments, and that various other changes and modifications may be made by one skilled in the art without departing from the scope or spirit of the invention.
