M. A.Gavrikov VISUALIZATION OF WINDOW INTERFACES IN "CONDOR" SYSTEM
The systems, containing knowledge, accumulated by some collective of people or by one person, are often dynamically changing. For example, the experience of the software developer, accumulated during his practice, is of the large significance.
Thus, there is the necessity of knowledge and experience presentation in such kind, which being sufficiently flexibile would optimally suit the storage of developing knowledge.
The example of the similar system could be the system of construction and visualisation of window interfaces - the Condor system.
The window interface is intended for logical linkage of different information units in one unit. The window interface represents a sequence of interconnected windows.
Such mechanism, as the dynamic creation of window in the environment of the system, was realized in this system. The internal structure of window is completely formed during work with program. It is realized with the help of dynamic menu call, where strucrure of a window will be sets with help of choice of accroding commands, and by the manipulation of window element prototypes.
This version of Condor system is realized on programming language Borland C++ 5.0 fore Windows 95, which allowed to use many of opportunities of this operating system and to create convenient user interface at creation of software.
The researches of the text information representation in the natural - professional language were carried out. The representation allows to make an operation of comparison of similar texts with each other, depending on their semantic contents.
Such kind of text matching is used in automated retrieval systems. The text at processing is represented as the frequent dictionary of words’s stems, excluding the stop words, i.e. the words not having a semantic importance. The retrieval system should make review of all search images, each of which represents the corresponding document, and compare them to the search instruction formed of the query text under the same laws and algorithms as search image of the document. While comparing the retrieval system calculates a coefficient of relevance (similarity) of the query to the document and thus has a capability of compilation of a list of the references to the documents ranked on a degree of relevance to the query. This list is query answer and represented to the user as an outcome of search.
For implementation of the retrieval system the algorithm of texts representation and comparing was developed. It rests on the correlation analysis, and also on the informational lingware (lists of stop words and morphemes of Russian). The program module realizing texts comparing and also the testing program were developed on the base of given algorithm. The testing program - retrieval system helped to research the qualitative side of the algorithm: time of search, noise and silence parameters at query answer.
The algorithm has shown a good outcome concerning triple of parameters speed - noise - silence: at text processing speed of 90 kbt/ss (Pentium-100) from each 8-th of the documents one will be lost only, on condition that on each relevant document one will be come noise.
Mulin A.S., Smirnov M.A. ATTEMPT OF USAGE OF THE CLIENT - SERVER TECHNOLOGY IN QUESTION-ANSWER PROCESSES
The version of the multi-user question-answer processor WIQA is realized.
At implementation the scheme of allocation of functions between server and client parts "the thin" server - "thick" client (for support of possibilities of single-user operation), also called as the two-layer client - server application is selected. For organization of the network protocol the Berkly sockets implementation for the set of operating systems Windows 9x/NT on the basis sets of protocols TCP/IP is used, that gives broad possibilities in operation of the system.
The principle « of the remote handle of the server » is applied. The server part of the application is realized with usage of possibilities of multithreading and synchronization of the application programming interface Win32. In it the streams adequate for operation of the interface with the user and for operation of the network interface with a client part of the application are divided. There is a user identification system on the part of the server on workstations.
Data remain deposited on-line on the server side. The administration of data access is realized on variant of pessimistic prediction, that is the occupied part of data temporarily is closed for changes on the part of other workstations up to it of release. For an uniqueness of displayed data the "upgrade service", tracking for changes of data in the system is realized. The principle of operation consists in following: application server watches changes of data, stored in it, and informs them to the clients, "interested" in changed data. In case of rise of fail situations the backup copying of used data is used.
For lowering the costs of resources of system data, used in calculations, are cached in memory of a workstation, also data compression, transmitted on the network is stipulated. During this version beta testing is defined, that quantity of maintained workstations basically is not limited, but hardly depends on the characteristics of the network platform from the set of operating system Windows 9x/NT. So at usage as the server platform of computer Celeron 450/ 64 Mb RAM / 4.3 Gb HDD under handle Windows 98 or Windows 98 Second Edition is provided operation from 30 up to 45 workstations depending from a kind of operation, thus in the system the course of operation above 10 projects is simultaneously recorded.
To improve the quality of engineering specifications it's appropriate to conduct a special processing of texts understanding them as "a primary meaning information" about reality. The version of predicative processing was taken for such processing.
The methods of predicative processing are investigated rather well and there's an experience in creating tools that are meant for predication.
A number of version of a tool LINA was worked out. In time these versions ceased to meet the demands of users and there appeared a necessity of making a new (a third under the account) version of LINA. The new version is written for the operational system Microsoft Windows 95/98/NT 4.0.
The offered envelope represents a toolkit meant for the formalization of statements of problems in SAP (System of Automatic Projecting) with the application of methods and means of making first order descriptions.
To reduce routine work, the new version has a mechanism of an automated analysis, that is, the selected sentence is analyzed according to one of the typical schemes.
In the process of this analysis the language data (the information about initial forms of words, grammar indication etc.) is recorded into the data base of LINA. To modify and correct this information the editor of language data LINA has been created. It has the following function possibilities: survey and editing of all data base. The work with data base is fulfilled in 2 modes: tables that represent information in a more visual traditional style and forms. The forms is a successive set of entries into the table or a number of tables, the variation of filter of the links of the table of main and dependents words according to the data: part of speech, member of a sentence installed separately for the main or the dependent word or, jointly, that is, there is a filtration of data base together with all possible combinations on the data: part of speech, member of a sentence, main or dependent word.
The program is designed as a multiwindow appendix. The main window has the main menu that allows the user to change the parameters of the set up of the appendix, to edit reference tables of grammar indications, parts of speech and members of the sentence. The main advantage of the appendix is a typical interface of access to data base.
For working it's enough to go the appropriate form and to start the survey, addition or editing of entries. There's a navigator in each window that lets move data base, add, delete and cancel editing.
After the analyses of the text the obtained predicates are recorded into data base forming a system of concepts represented as a semantic web. Each entry in data base corresponds to a certain quantum of knowledge with appropriate applications and attributes. There was offered the following structure of data base to store this information. The main thing is a concept that has the following attributes: rang, contents, expression and application. The attribute of the application is taken into a separate essence as it has a number of basic attributes: condition and model. Besides, each of these attributes is composite since it combines such attributes as contents and expression.
Such model of data lets represent the information about data base in more adequate way.
The editor of concepts has a possibility of surveying, editing, addition of separate quanta of knowledge and applications. The system of search allows to find this or that concept with a certain set of properties. Each property is displayed on the screen in an appropriate field which gives an opportunity to represent the information about the concept visually.
The appendix were worked out for Windows that's why the interfaces are easy to understand as well as all the interfaces of appendices worked out for the operational system Windows 95/98/ NT 4.0.