HCS: Study On Algorithms And Models Of Decision Making Problem Based On "Human-Centered Service"

In the existing decision-making support systems, humans are oftentimes the submissive recipients,and thus, the results produced by these systems sometimes differ significantly with the users’ idealdecision-making scheme, which greatly impacts the reliability and reputation of these systems. Ashuman is a key element in improving the effect and efficiency of decision making problems, wepropose a “human-centered service” model for complex decision-making problems, highlighting thathuman is also an indispensable “decision-making factor” and human preferences could also affect thedecision-making results considerably. The goal of studying on methods and models based on“human-Centered Service”(HCS) decision making problems is to build a complete system conceptabout computer-aided human centric decision support system, which can provide advanced methodsand models support for the development of human-machine coordination decision support system.The decision making problem with human centered service is one of the innovation of the forms intraditional decision problem.“Human-Centered Service” decision making problems are oneinnovation and breakthrough for traditional decision problem. The research primarily focuses on themethod and model of decision making problems based on the HCS. To improve the human servicesfunctions and reliability, this paper primarily focuses on several significant aspects, including designprocesses of decision-making problem recognition, task-driven decision decomposition strategies andthe mutual transformation methods among these different strategies, decision-making schemeadjustment algorithms and effect assessment models based on decision-makers psychologicalthreshold, and decision-makers psychological expectation rules and information fusion methods basedon decision-makers satisfaction, etc. The new methods and models break through the existingarchitecture for decision support system, and provide more efficient auxiliary decision.In this paper the main innovative research results are shown as follows:(1) Study on Criteria Weights for Decision Making Problems Based on “Human-Centered Service”.This paper mainly focuses on the uncertainty of criteria weights problem. In this section, we firstpropose two new methods to obtain the criteria weights based on the similarity relation andadvantageous relation, respectively. Then, we points out the similarity relationship and advantageousrelationship between the decision making alternative(s) and the ideal decision alternative has therelation of equivalence with the probability measure of criteria value. Later, with the help ofmaximizing deviation algorithm rules, we propose the criteria weights based on the similarity andadvantageous of criteria values, then following the similarity and advantageous between decision making alternative and the ideal standard, we rank and pick over all decision making alternatives.(2) Study on Alternatives Ranking for Decision Making Problems Based on “Human-CenteredService”. In this section, first, we put forward transformation methods for interval number intoconnection number, and use the connection number propose a new possibility degree formula forinterval numbers. Then, we evaluate the magnitude of interval number for decision making advantagematrix according to its possibility degree. Later, we research on the chanellege that different intervalnumbers have the same possibility degree values. Third, we divide the decision makers into threedifferent types according to their risk preferences and we proposed three risk preferences assumptionmodels corresponding to different type decision makers to determine the advantage relation forinterval numbers. Finally, we propose advantage relation, advantage degree of the alternatives,advantage matrix to rank all the alternatives, and find out the best one based on the weightedcombinatorial advantage values (WCAV) of alternatives.(3) Study on Decomposition Strategies ans Criteria Reduction in Multi-Criteria Decision Making(MCDM) Problems Based on“Human-Centered Service”. In this paper, we propose two differentstrategies to address the “large decision table”(e.g. a large number of criteria) challenge in multiplecriteria decision making with interval number. One strategy: we first classify decision makers (DMs)into three types according to their risk preferences, then we use different methods to find usefulcriteria, obtain criteria weight for useful criteria, information fusion, and rank alternatives accordingto weighted combinatorial advantage values (WCAV) corresponding to different types DMs, then weselect the most desirable choice(s) for different risk preference decision makers. Another strategy: wefirst find out the useful criteria and obtain criteria depending on the data. Then, using the differentmethod to rank and select the most desirable choice(s) corresponding to different types DMs.(4) Study on Revisiting and Reliability Targets for Multiple Criteria Decision Making ProblemBased on “Human-Centered Service”. The idea in this section is that the decision maker’s utility orvalue may not depend on the levels of performance on different criteria, but instead on whether thelevels meet a target or threshold on one or more criteria. First, we build the expectation functions ofdecision makers for three different type criteria. Then we filter off the useless alternatives and build anew decision table that can reach to the basic satisfaction degree on those criteria that decisionmaker’s with an expection vaule or interval. Second, we propose a new method to obtain criteriaweights using the satisfaction degree of the decision makers. Third, we also propose a new definitionfor satisfaction degree of alternatives. The best alternative(s) in decision making found by thetraditional ranking methods does not have an equivalence relation with the alternative(s) that have thebiggest combinatorial satisfaction degree of decision makers.(5) Study on Decision Processes Abstraction and Optimization Based on Process Mining for “Human-Centered Service”. In this section, we use adjacency matrix to extract3types of annotationsinformation (i.e., frequencies, remaining times, activities) and use them to construct new processmodels from the event logs. First, a new process model based on adjacency matrix is proposed.Second, by adding the stage and frequency for every activity into the matrix, another new processmodel based on stage adjacency matrix is further proposed to avoid the possible loops or self-loops.Third, based on the second new model with a multi-stage structure, we compute the conditionalprobability from every stage to next stage through the frequency. These new process models can beused to predict what will actually happen, how possible to reach the next activity, and how soon forthe ongoing process instances to finish. Later, another two handover process models are extractedbased on person adjacency matrix abstraction such as how work transfer from one person to another.The new handover process models can be used to optimize the handover business process throughprobability and time prediction. We use the decision process models to analyze decision maker’spreferences and improve the human preference models.