Of course the ScienSe's development team is conscious about the weak point of the product, thus thinking about them, we present the following list of characteristics that could help to eradicate of such weaknesses.
- The first thing considered of vital importance is reinforcing the extracting information algorithm, since the one that is implemented in this version, present some fails in front of certain articles with outside the ordinary format, because it is based on probabilities. We have thought that an alternative to this algorithm is to use a trained neural network, since they have demonstrated in numerous occasions ideal characteristic for this kind of work.
- Another improvement that could hit in a significant way the usability of the system is the creation of a friendly graphical user interface for the configuration of the indexer since, in this moment, this process is through a configuration file that, in spice of being of easy creation and modification, forces the end user to submerge in a sea of information that would be not so useful for him, even more when he requires to make things as easy as add a new url to the searching. The creation of an interface that contained only the variables that the user should configure and which can be able to add those values to the indexer's configuration file, would allow an even more easier interaction of the user with the program. At the present time we have a version of such application, called htconf but it still in a development phase. It would also be desirable to allow the configuration of the system through the Web.
- Another one of the things that are being developed is JDBC Tester, a tool to prove JDBC connections and SQL statements, by means of a friendly user interface. At the present time it is already functional but it still need some improvements in the presentation of the results and in the system's help.
- Adding the possibility of doing combined searches is another one of the improvements that are consider, for this it would be necessary to modify the components that access to the database and to add code to the interface, so that facilitate to the user's the searching process.
- When htsciense finds files with the formats: PDF, HTML, DOC, PS, etc., creates a process that calls a filter that makes the conversion of the given format into plain text. These filters need to be improved and to developed more so as to handle other formats.
- In the future we will find that for different types from publications certain extracting information algorithms are better than others, in addition of there will be who want to prove a new algorithm without need to modify all the source code of htsciense. Therefore we believe that is important to add the plug-ins support in order to allow the use of different algorithms with neither any needed of compiled the application each time nor knowing the details of its operation.
- In order to improve the user's participation, we think that is important give them the possibility of modifying the information of their own publications at any moment, as well as to add new data. In such a way that each author can participate actively in the getting information process so as to make the system more valuable. Therefore, adding user's sessions would allow to improve as the interface as the trustworthiness of ScienSe.
- In order to allow that ScienSe can adapt better to the requirements of different users without having to invest long time and resources in adapting the system to the particular necessities, we want to introduce in ScienSe the Model-View/Controller in the application, in order to separate the logic from the presentation. In this way the modification of the interface won't require of programming knowledge.
- The use of frames was for improving the user interface. But later we realized that it could have problems in those browsers without frames support, therefore we believe advisable to modify the application so that it does not use frames.
- There exist some data that can be useful to know about a scientific publication, but even though that information can already be obtained from the system, there is no a place within the interface that shows this results to the user. Between the data that could be of interest there are: the most referenced authors, the most referenced publications. Thus a possible improvement, would be to add a section in the web interface which shows this information.
- Once it has the sessions' support, it can be able that the users could choose between different web interfaces, depending on which is the one that better adapts to his necessities, this can be done by means of themes support in the interface.
- Given the few resources with which we made the project, there was not possible to test the system in different platforms, therefore it would be desirable to prove the program in different platforms of operating system like MacOS X, FreeBSD, Solaris, Debian, Slackware, for mentioning some. Besides to prove the system in different implementations of J2EE, in other database manager systems as MySQL, Oracle, Firebird. And to elaborate packages for the different platforms.
- Another one of the improvements that can make of ScienSe a better tool for working and could make easier to get its maximum benefit would be to add support for JavaHelp, so as to have a powerful and consistent help system.
- We wish to improve the application logo also, thus we hoped that some user with knowledge of graphical design can help us. The unique requisite is to make it using The Gimp.
Search By Google