]Help: 1. level: FAQ
The FAQ, which are the answers for those individual questions that are frequently asked by the users, sums up the knowledge of those questions that may possibly interest many other users: (Demo)
1. question: How can one get a rapid overview of the steps required for initialization of the analytical modules?
1. answer: The modules are found in the list at the sitemap. The introduction sites of the modules include the list of the prerequisites of the program execution. Also, there is a demo file that shows a specific (reproducible) example, which should make the creation of one's own task self-evident...
2. question: Is the webpage only for education purposes, or it can be used by anyone?
2. answer: The primary aim of MY-X FREE is education, but of course everyone who found it can use it.
3. question: When would the system go live (not learning)?
3. answer: Some elements of the system may already be considered alive (e.g. algorithms of the limitless executions run as final versions). A so called live system may never be created as a unique service, instead of that, special target groups may find the personalized versions of the services shown here.
4. question: From when, and how much shall be paid for the use of the system?
4. answer: At the moment, there is no pricing for the tasks beyond the limit or for other services (e.g. online expert systems, creation and upkeep of OLAP-databases). The motivations of the partners should be clarified by individual contact. In case of interesting applications, the reference-making effect shall be considered by the partners too...
5. question: What is an OAM, and what do the other abbreviations stand for?
5. answer: OAM = the object-attribute matrix, the learning pattern that the offered algorithms learn from. CBR = case-based reasoning: The collective term for those procedures that are able to learn and reason based on an OAM. COCO = component-based object comparison for objectivity STEP (STD vagy IX) = a similarity analysis that consists of multiple steps, which aims to, for example, grope the mInimum or maXimum (cf. IX) balanced prices, or the involvement of the unused attributes in the model. MCM = Monte Carlo Method, a totally or partially random search. OLAP = online analytical processing, a complex real time database-query...
6. question: Does the usage of the system require any mathematics/informatics knowledge?
6. answer: The MY-X FREE does not need any knowledge other than reading, because the sites appearing in the browser does not expect any more than few clicks and the specification of few characters.
HOWEVER: Not understanding the mathematical foundations of similarity analysis affects the usage of the system fundamentally. The first steps of learning shall be taken where the OAM that someone wants to reinterpret already exists. So the OAM is already at the desk of someone, and based on the OAM, someone calculates or thinks something...
The MY-X FREE aims to partially automate the methodology of the conclusions drawn from past experiences! The offline similarity analysis assumes the knowledge of the RANKING(), VLOOKUP(), SUM and IF() MS-Excel functions, and the SOLVER module.
7. question: What are the technical requirements for running the system?
7. answer: The usage of MY-X FREE does not have any special prerequisites: the HTML-pages should able to be loaded in all known browser of all operating systems. It is necessary to allow javascript to run, and to allow the opening of POP-UP windows and messages, and to view the website in different languages, one shall know about the language options of the browser. The MY-X FREE handles each step on server side. It gets the inputs from the browser, and returns the calculation results in it. In order to use the results in a spreadsheet software (like MS-Excel) without any problem, change the list separator to dot instead of comma in the regional and language settings.
8. question: What shall be done if an OLAP-query returns with no results?
8. answer: OLAP databases do not necessarily have data for all available options of the query interface. Especially if, for example, we expect that temperature-data is given in %. (Multiple warnings may be the result of the logic of multi-component query.) In such cases the reality of the query shall be re-evaluated. If there is no error in it, then the required data does not exist or it is not yet stored.
9. question: What are the main differences between the analytical modules?
9. answer: The modules offered here differ in the search conditions of the stairs (that mean the solution). The modules can be ordered, for example, by the strictness of their search conditions (the first module in the order leads most likely to zero model error, whereas the last module has much more errors): the MCM module uses the least expectations (quasi no restriction), which gropes the potential ceteris paribus figures (monotone growing, shrinking, optimum-like) with a small number of stairs. It is followed by the OPTI_2 module, which expects optimum-like effects for certain elements of the stair-pairs. The OPTI_1 module provides the resultant of the ceteris paribus figures alone, or more likely, system-wide. The COCO STD module (contrary to the permissive, explanation-seeking settings) means the average strictness level in the similarity analysis, because of its monotone direction conditions. The COC_S module limits the attributes maximum effect, and the distance of any two stairs, therefore it means stricter conditions than the standard level. The Y0-model does not fit to this series, because in its case, the Y data vector is missing from the OAM, so it expects a special starting situation, which is incomparable with the other methods. The STEP STD and STEP IX (just as it can immediately be seen in its menu) are not separate modules, they are step sequences that are based on the other modules (primarily on COCO-STD).
10. question: Can an OAM be run on any analytical modules?
10. answer: Theoretically speaking, yes. However, the results may need to be interpreted differently, based on the logic of the 9. question. Among those modules that need dual input signal, there is the COC_S module which equals a COCO STD module, if we give a single input signal to it. The OPTI_1 module, which needs a dual input signal too, will lose its monotonity-breaking effect with a simple OAM, because then there will be no such data in the model that has an inverse meaning.
11. question: Can the analytical modules be combined?
11. answer: Yes! Moreover, the production of a consistent result (cf. forecast, and a rule set that explains how a system works) demands the evaluation of the same data asset from different points of view. Of course, the revelation of inconsistency can not only be reached through module-switching, but with the alteration of the OAM's content too: (cf. consistency-figures).
12. question: In what kind of framework was the similarity analysis tested so in case of corporate/institute integration, what kind of alternatives can be chosen from?
12. answer: Similarity analyses can be done in offline environment, using in-built solver of spreadsheet software (cf. MS Office Excel), or one's own macros (Open Office was not yet tested successful). Besides that, server-side php-programs can be used as online solution (take MY-X FREE for example). Online environment was simulated on e.g. Windows XP (cf. apache/postgresql/php), which solution didn't need internet connection. A virtual machine (cf. linux/apache/postgresql/php) installed on windows vista works too, but it requires DHCP service. Among the future plans, there is the examination of moving the similarity analysis calculations to client-side, the definition of large-scale analyses (cf. SETI) as tasks that require shared resources, and the foundation of the service to send data and receive results on cell phone.
13. question: How does online similarity analysis fit to the services of an education/research institute?
13. answer: The applicability of similarity analysis is the matter of the organization level of the institute's online processes. If an institute has modules that are operable one their own, then the similarity analysis can be considered as a standalone module, which does not make any conflicts with other running applications, be it operated internally or be it outsourced. If the institute actively faces consolidation questions (e.g. unified handling of student news, merging of data assets like time tables, quality assurance of mashup-like data collections), then OLAP solutions, that are the basis of similarity analysis, are hard to avoid. If the consolidation is successful, then it raises the question of integration of the different information sources (e.g. joint news databases, LDAP, joint research databases, joint library systems). The fourth level of development is optimization, which can be considered as some sort of a cloud computing, into which outsourcing solutions fit too. Optimization assumes the SOA and BPM like planning, so that level of granularity, which does not affect the basic modules (e.g. transaction databases) when the organization needs to react to the business model changes on the interface. The fifth level is the innovativeness, so the involvement of similarity analysis in the initialization of the organization's strategy (revelation of development directions, recognition of balance losses, making sustainability forecasts). Therefore, MY-X FREE represents the entirety of online services both from the technology and content point of view.
14. question: What does the online similarity analysis has to offer in regard of the information-evolution model?
14. answer: Infrastructure: The MY-X FREE means a centrally developed and maintained infrastructure that can be placed into the integration processes almost arbitrarily. Knowledge process: The OLAP-based similarity analysis produces online expert systems that embody the fact-driven modeling with almost unlimited automatability. Human capital: MY-X FREE offers users to spend their intuition driven skills on creation (modeling), thus they don't have to deal with repetitive (automatable) tasks. In case of theses and PHD theses, it is none other than the expectation to exceed the best practice and to make source code out of it - instead of descriptive, only text-producing pseudo-solutions. MY-X expects those who get involved with it to put all steps under the magnifying glass critically, keeping the principles of quality assurance in mind, from the planning of the data assets to the interpretation of the results. Culture: The routine operation of the aforementioned parts that are closely related to each other is the XXI century work culture, which relies on the correct and public maintenance of data assets, expects high level of innovativeness of its elements while the framework is under constant balance checks - thanks to the similarity analyses that offer the automation of human capabilities...
15. question: What is the difference between COCO online MCM and COCO offline MCM?
15. answer: In case of the COCO online MCM module, the illusion of flexible searching is generated by not giving of directions for the attributes, so the MCM module is such an optimization where the minimization of the model's error makes parameters out of the directions of the attributes, because the small number of stairs (default = 3) can have arbitrary figures. In the offline MCM, flexible searching implies the search for the stairs (that are the solutions of the similarity analysis) by guessing (RND), which means that in the spreadsheet software, the values of the first stair row have to be placed between 0 and the maximum of the Y-vector, while all other stairs have to placed between the stairs above them and 0 (or the negative Y (max)). This way, based on the MCM's logic, with many tries, occasional OAM antagonisms can be approximated well.
16. question: What kind of antagonisms exists in regard of similarity analysis?
16. answer: Direct antagonisms are those cases, where there are at least 2 objects (rows) in an OAM, from which one has no worse rankings than the other, but for one of them, the consequence (e.g. price) is better.
Indirect antagonisms are there in those OAMs that have no direct antagonism, but a zero-error stair function cannot be fitted during the modeling, so the interference is not in the different input-logics of the two objects, but there is a complex contradiction between 'n' objects.
17. question: ?
17. answer: .
