NTEGRATION OF AUTOMATIC LEARNING AND PLANNING FOR PRESCRIPTIVE ANALYTIC. APPLICATION TO DECISION-MAKING IN TRANSPORT COMPANIES (Q3136267)

From EU Knowledge Graph
Revision as of 08:21, 9 October 2021 by DG Regio (talk | contribs) (‎Changed an Item: Edited by the infer coords bot - inferring coordinates from location)
Jump to navigation Jump to search
Project Q3136267 in Spain
Language Label Description Also known as
English
NTEGRATION OF AUTOMATIC LEARNING AND PLANNING FOR PRESCRIPTIVE ANALYTIC. APPLICATION TO DECISION-MAKING IN TRANSPORT COMPANIES
Project Q3136267 in Spain

    Statements

    0 references
    30,782.4 Euro
    0 references
    38,478.0 Euro
    0 references
    80.0 percent
    0 references
    1 January 2019
    0 references
    31 December 2021
    0 references
    UNIVERSIDAD DE GRANADA
    0 references

    37°10'24.60"N, 3°35'58.31"W
    0 references
    18087
    0 references
    LA ANALITICA PRESCRIPTIVA ES LA ULTIMA FASE DE LA ANALITICA DE NEGOCIOS, QUE INCLUYE COMO PASOS PREVIOS LA ANALITICA DESCRIPTIVA Y LA ANALITICA PREDICTIVA. LA ANALITICA PRESCRIPTIVA REQUIERE LA APLICACION DE DIVERSAS TECNICAS PARA APROVECHAR LOS RESULTADOS DE LA ANALITICA DESCRIPTIVA Y PREDICTIVA. ESTAS TECNICAS PUEDEN INCLUIR TANTO EL APRENDIZAJE AUTOMATICO COMO LA PLANIFICACION AUTOMATICA. _x000D_ _x000D_ EN NUESTRO CASO, APLICAREMOS EL ANALISIS PRESCRIPTIVO AL AMBITO DEL TRANSPORTE INTELIGENTE. PARA ELLO CONTAMOS CON LA COLABORACION DE UNA START-UP TECNOLOGICA (GANTABI). GANTABI ES UNA PLATAFORMA BASADA EN LA NUBE QUE CONECTA A LOS PROVEEDORES DE SOFTWARE NACIONALES Y MULTINACIONALES EN LAS VERTICALES DE TRANSPORTE Y LOGISTICA._x000D_ _x000D_ LA EMPRESA DISPONE DE UNA GRAN CANTIDAD DE DATOS RELATIVOS A LA OPERACION DE LOS DISTINTOS CLIENTES QUE PRESTAN SERVICIOS, Y NUESTRO OBJETIVO ES OFRECER RECOMENDACIONES SOBRE LAS ACCIONES QUE DEBEN LLEVARSE A CABO PARA ALCANZAR LOS OBJETIVOS DE NEGOCIO. PARA ELLO SE UTILIZARAN LOS DATOS DE LA ACTIVIDAD ANTERIOR REALIZADA POR LA EMPRESA Y REGISTRADA COMO LOGS DE ACTIVIDAD. _x000D_ _x000D_ ESTE OBJETIVO NOS PERMITIRA APOYAR LAS DECISIONES DE LA EMPRESA, RECOMENDANDO UN CURSO DE ACCION QUE LE PERMITIRA ALCANZAR LOS OBJETIVOS DE NEGOCIO. PARA LOGRAR ESTE OBJETIVO SE INTEGRARAN TECNICAS DE APRENDIZAJE AUTOMATICO Y DE PLANIFICACION AUTOMATICA, BASADAS EN LOS RESULTADOS OBTENIDOS EN EL PROYECTO ANTERIOR (TIN2015-71618-R). EN ESTE PROYECTO EL EQUIPO DE INVESTIGACION HA INICIADO UNA NUEVA LINEA DE INVESTIGACION SOBRE LA PLANIFICACION DEL APRENDIZAJE EN EL DOMINIO. LAS TECNICAS DESARROLLADAS PERMITEN GENERAR DOMINIOS DE PLANIFICACION HTN A PARTIR DE TRAZAS DE PLANES, ASI COMO DOMINIOS DE PLANIFICACION PDDL CON INFORMACION LOGICA Y NUMERICA. ESTAS TECNICAS SE ESTAN APLICANDO A PROBLEMAS DE TRANSPORTE INTELIGENTE DESDE NUESTRA COLABORACION CON LA EMPRESA GANTABI._x000D_ _x000D_ POR LO TANTO, NUESTRO PROPOSITO ES PROPORCIONAR UNA SOLUCION AL PROBLEMA DE LA ANALITICA PRESCRIPTIVA MEDIANTE LA AMPLIACION DEL PROCESO DE APRENDIZAJE AUTOMATIZADO YA DESARROLLADO CON EL FIN DE AUMENTAR LA EXPRESIVIDAD DE LOS DOMINIOS APRENDIDOS QUE ABORDAN LOS SIGUIENTES TEMAS: (1) ENRIQUECER LOS DOMINIOS DE PLANIFICACION CON CONOCIMIENTOS ADICIONALES RECUPERADOS Y EXTRAIDOS MEDIANTE EL ANALISIS Y LA EXTRACCION DE REGISTROS DE ACTIVIDAD CON INFORMACION CONTEXTUAL (POR EJEMPLO, MEDIANTE LA CARACTERIZACION AUTOMATICA DE OBJETOS Y RECURSOS QUE INTERVIENEN EN EL PROBLEMA DE PLANIFICACION) (2) LA GESTION DE LA INCERTIDUMBRE, APOYANDOSE EN LENGUAJES DE DOMINIO DE PLANIFICACION ESTANDAR BASADOS EN LA PROBABILIDAD BASADOS EN LA HTN O EN LA PRIMITIVA (3) OPERACIONALIZAR LOS DOMINIOS APRENDIDOS PARA QUE PROPORCIONEN SERVICIOS DE ANALISIS PRESCRIPTIVO MEDIANTE LA REALIZACION DE UN PROCESO HIPOTETICO QUE INTEGRE LA PLANIFICACION Y LAS TECNICAS DE APRENDIZAJE AUTOMATICO. (Spanish)
    0 references
    PRESCRIPTIVE ANALYTICS IS THE LAST PHASE OF BUSINESS ANALYTICS, WHICH INCLUDES AS PREVIOUS STEPS DESCRIPTIVE AND PREDICTIVE ANALYTICS. PRESCRIPTIVE ANALYTICS REQUIRES THE APPLICATION OF VARIOUS TECHNIQUES TO TAKE ADVANTAGE OF THE RESULTS OF DESCRIPTIVE AND PREDICTIVE ANALYTICS. THESE TECHNIQUES MAY INCLUDE BOTH MACHINE LEARNING AND AUTOMATIC PLANNING. _x000D_ _x000D_ IN OUR CASE, WE WILL APPLY THE PRESCRIPTIVE ANALYSIS TO THE FIELD OF INTELLIGENT TRANSPORT. FOR THIS WE COUNT ON THE COLLABORATION OF A TECHNOLOGICAL START-UP (GANTABI). GANTABI IS A CLOUD-BASED PLATFORM CONNECTING TO NATIONAL AND MULTINATIONAL SOFTWARE VENDORS IN TRANSPORT AND LOGISTICS VERTICALS._x000D_ _x000D_ THE COMPANY HAS A LARGE AMOUNT OF DATA RELATING TO THE OPERATION OF THE VARIOUS CUSTOMERS WHO PROVIDE SERVICES, AND OUR OBJECTIVE IS TO PROVIDE RECOMMENDATIONS ABOUT THE ACTIONS THAT SHOULD BE TAKEN TO ACHIEVE THE BUSINESS GOALS. FOR THIS PURPOSE, THE DATA OF THE PREVIOUS ACTIVITY CARRIED OUT BY THE COMPANY AND RECORDED AS LOGS OF ACTIVITY, WILL BE USED. _x000D_ _x000D_ THIS OBJECTIVE WILL ALLOW US TO SUPPORT THE COMPANY'S DECISIONS, RECOMMENDING A COURSE OF ACTIONS THAT WILL ALLOW YOU TO ACHIEVE THE BUSINESS GOALS. TO ACHIEVE THIS GOAL WE WILL INTEGRATE MACHINE LEARNING AND AUTOMATIC PLANNING TECHNIQUES, BASED ON THE RESULTS OBTAINED IN THE PREVIOUS PROJECT (TIN2015-71618-R). IN THIS PROJECT THE RESEARCH TEAM HAS STARTED A NEW RESEARCH LINE ABOUT PLANNING DOMAIN LEARNING. THE DEVELOPED TECHNIQUES ALLOW FOR GENERATING HTN PLANNING DOMAINS FROM PLAN TRACES AS WELL AS PDDL PLANNING DOMAINS WITH LOGICAL AND NUMERICAL INFORMATION. THESE TECHNIQUES ARE BEING APPLIED TO PROBLEMS OF INTELLIGENT TRANSPORT FROM OUR COLLABORATION WITH THE COMPANY GANTABI._x000D_ _x000D_ THUS, OUR PURPOSE IS TO PROVIDE A SOLUTION TO THE PRESCRIPTIVE ANALYTICS PROBLEM BY EXTENDING THE AUTOMATED LEARNING PROCESS ALREADY DEVELOPED IN ORDER TO INCREASE THE EXPRESSIVENESS OF THE LEARNED DOMAINS ADDRESSING THE FOLLOWING ISSUES: (1) ENRICHING PLANNING DOMAINS WITH ADDITIONAL KNOWLEDGE RETRIEVED AND EXTRACTED BY ANALYZING AND MINING ACTIVITY LOGS WITH CONTEXTUAL INFORMATION (E.G. BY AUTOMATICALLY CHARACTERIZING OBJECTS AND RESOURCES THAT INTERVENE IN THE PLANNING PROBLEM) (2) MANAGEMENT OF UNCERTAINTY, RELYING ON STANDARD PLANNING DOMAIN LANGUAGES BASED ON PROBABILITY EITHER HTN-BASED OR PRIMITIVE-BASED (3) OPERATIONALIZE THE LEARNED DOMAINS TO PROVIDE PRESCRIPTIVE ANALYTICS SERVICES BY CARRYING OUT A WHAT-IF PROCESS THAT INTEGRATES PLANNING AND MACHINE LEARNING TECHNIQUES. (English)
    0 references
    Granada
    0 references

    Identifiers

    RTI2018-098460-B-I00
    0 references