Support for Graphical Modelling in Bayesian Network Knowledge Engineering: a visual Tool for Domain Experts




НазваниеSupport for Graphical Modelling in Bayesian Network Knowledge Engineering: a visual Tool for Domain Experts
страница9/35
Дата05.10.2012
Размер0.88 Mb.
ТипДокументы
1   ...   5   6   7   8   9   10   11   12   ...   35

2.4Existing tools for building Bayesian Networks


The stage of designing a BN must be followed by the process of constructing the Net, i.e. entering the data of the model into the computer inference program. Most BN software packages use graphical user interface and standard windows interface i.e. windows, menus, mouse clicks and drag-and-drop, for creating, editing and running BN models. The main display focus in these packages is the graph structure. Examples for such packages are HUGIN [49], Netica [66], Analytica [5], GENIE [29] and MSBNx [63]. All these packages use:


Variables – an oval object in the program represents each variable in the model, which is represented by a node in the graph. The node object contains the node properties. These include the name, title, description, and probabilities of the node. In some packages, such as Analitica and GENIE, it is possible to define sub-models, represent them by a special node and then zoom in and out. This capability is very useful for the design of large networks.


Relationships – the direct relationships between the variables is represented by directed arcs between the corresponding nodes. Nodes and arcs are easily created and can be edited with mouse clicks and mouse drag-and-drop.

Parameters – The parameters are specified locally. A conditional probability table (CPT) for discrete probability distributions or a function for non-discrete probability distributions is attached to each node. The user is provided with an interface to define the quantitative relationships between the node and its parents. In the case of CPT this interface is a table in which each cell contains the probability of one state of the variables and a certain combination of the parents states.

  1. Representation of the CPTs

There are various existing methods to support the user in filling-up these tables. In HUGIN and GENIE the rows are the parents states and the columns are the node states. Netica represents a transposed matrix of this table. In all packages if the table is larger than the screen the user is provided with scrolling bars. In some BN software packages (e.g. MSBN) only one CPT column is visible at a time; in some the conditional probabilities and the parent combinations list are displayed as separate tables; and in some as a probability tree [101].



  1. Supporting the elicitation process

Although there are various methods to support the elicitation process of parameters (see section 2.3.3), only the probability wheel is implemented in the general BN packages. For example, this method is implemented as a pie chart in MSBN, and as a pie chart as well as a bar graph in GENIE [101].


  1. Interactive assessment of parameters

SAMIAM [14] enables the user to perform Constraints Enforcement analysis and Analytica and Netica offer Importance analysis. In Netica the user has to choose a node of interest and the analysis displays all nodes that can influence the chosen node ordered by the strength of influence. In Analytica, the user can ask for Importance analysis for a selected node of interest. The output is a report that displays how much the states of the query node could be influenced by a single finding in each of the other nodes in the network. HUGIN enables a What-If analysis and Conflict Analysis.


Assessment

Case-Based assessment: This assessment is possible in all tools. Scenarios can easily be entered using the graph display and the change in probabilities can be easily read from the graph.


General-View assessment: This assessment needs the support of programming experts or BN experts in order to manipulate the packages’ Application Programmer Interfaces (API). The ability to compare the results obtained from the model (the posterior probabilities or any function thereof) with the experts’ expectations is partially supported in Netica where a file containing cases can be supplied by the user, Netica then reads the file and compares the cases with the results of the model. It accumulates all the comparisons and produces a summery statistics. The user cannot change the output and cannot get additional information with regard to specific cases without using programming skills and using the API.


Iterative Process

Most tools are designed to support the construction of the Net, and therefore applying changes to the network is accordingly easy. As a consequence of this, these tools also support the iterative nature of BN construction. However, none of the tools provides a direct functionality for version control of the constructed networks.

1   ...   5   6   7   8   9   10   11   12   ...   35

Похожие:

Support for Graphical Modelling in Bayesian Network Knowledge Engineering: a visual Tool for Domain Experts iconThis document is being distributed by the American International Health Alliance (aiha) / the EurasiaHealth Knowledge Network with support from the us agency
Саратовский областной Центр по профилактике и борьбе со спид и инфекционными заболеваниями
Support for Graphical Modelling in Bayesian Network Knowledge Engineering: a visual Tool for Domain Experts iconStudents will demonstrate competence in; Engineering Curriculum; basic concepts such as significant digits, length, and mass; Engineering Drawing; the use of the computer as an engineering tool; designing a project

Support for Graphical Modelling in Bayesian Network Knowledge Engineering: a visual Tool for Domain Experts iconAn Object-Oriented Support Tool for the Design of Casting Procedures

Support for Graphical Modelling in Bayesian Network Knowledge Engineering: a visual Tool for Domain Experts iconRegional network of co-operation in Europe as a tool of efficient governance

Support for Graphical Modelling in Bayesian Network Knowledge Engineering: a visual Tool for Domain Experts iconEnabling Decision Support and Costing of Product Designs by using Visual Metaphors

Support for Graphical Modelling in Bayesian Network Knowledge Engineering: a visual Tool for Domain Experts iconRepresenting and using legal knowledge in integrated decision support systems: DataLex WorkStations

Support for Graphical Modelling in Bayesian Network Knowledge Engineering: a visual Tool for Domain Experts iconThis document contains a list of references to publications and reports about Bayesian Net technology, and especially Bayesian Net applications. The report will

Support for Graphical Modelling in Bayesian Network Knowledge Engineering: a visual Tool for Domain Experts iconNorwegian Network for the Support of the Indigenous Peoples of the Russian Arctic (nnsipra) Сеть Норвежских Организаций в Поддержку Коренных Народов Российского Севера

Support for Graphical Modelling in Bayesian Network Knowledge Engineering: a visual Tool for Domain Experts iconDevelopment of a Product model to support engineering change management

Support for Graphical Modelling in Bayesian Network Knowledge Engineering: a visual Tool for Domain Experts iconTool Use: Misuse of shop tools will result in the loss of tool privileges

Разместите кнопку на своём сайте:
Библиотека


База данных защищена авторским правом ©lib.znate.ru 2014
обратиться к администрации
Библиотека
Главная страница