o:31954 Модел за детекцију и анализу узрока кашњења на пројектима базиран на подацима издвојеним из неструктурираних извора докторска дисертација A model for detection and analysis of causes of delay on projects based on unstructured textual sources : doctoral dissertation sr кашњење, базни узроци кашњења, изградња путнеинфраструктуре, машинско учење, рударење по текстуалним документима,Transformer, неструктурирани подаци Time overrun in construction projects is a global phenomenon that has beenresearched for decades. The traditional approach to detection and analysis ofcauses of delay usually involves gathering experts’ experiences acquired on similarprojects (grouped by their type or geographical location). The result of such anapproach is list of causes of delay, hierarchically arranged according to theirimportance. Such empirical research is burdened with bias and subjectivism ofexperts and does not lead to the detection of the root causes of delay at a singleproject level.A database, formed using data collected from 75 road infrastructure projectsimplemented in Serbia between 2004 and 2021, is used to demonstrate thetraditional approach’s weaknesses and to create a basis for establishing a newapproach. The results of research and analysis of the database show that over 80%of projects got delayed with an average time overrun greater than 90% of thecontract duration. Based on the survey involving key stakeholders on the databaseprojects, a causes of delay list was formed that does not deviate from lists in moststudies. Furthermore, low values of Spearman rank correlation were obtained(0,204 - 0,565) between attitudes of different stakeholders, which confirms thesignificant presence of subjectivism and bias in the conducted empirical research(surveys).The main goal of the doctoral dissertation is to create a new model for unbiaseddiscovery of the root causes of delays at the single project level and its entities,using machine learning on unstructured text documentation from the project. Thechosen documentation for the development of the model is Minutes of Meetings(MoM) because they contain comprehensive information about delays, whichoccurred at the time of the issues, with a precise time frame. Machine learningtechniques using Transformer language models enable automatic detection ofcauses of delays. Focused expert knowledge is used for additional unbiasedtraining of the model for the selected domain of road infrastructure, by connectingparts of the text with causes of delay from a previously defined list. Recognizedentities of road infrastructure projects are tunnel, route, and bridge. By combiningthe mentioned elements, the dissertation developed an analytical Model for thedetection and analysis of the causes of delays in road infrastructure constructionprojects, called DREAM (Delay Root-causes Extraction and Analysis Model).In the first phase, DREAM automatically generates a causes of delay list by projectentities, based on the frequency of their occurrence in Minute of Meetings. Theresults show that the model can detect the causes of delay, returning acceptablerecall values (recall = 0.69, for the most frequent causes of delay).In the second phase, enabled by MoM dates, DREAM adds a new and uniquefeature – graphs of the temporal distribution of causes of delay during the project.xBy qualitatively analyzing these graphs that show the frequency and intensity ofindividual causes of delay, experts can understand the nature of the causes ofdelay, which enables them to detect root causes, the ultimate goal of all researchrelated to delays on construction projects.The conducted research provides scientific and practical contribution. A newapproach to the causes of delays identification and analysis is proposed through adeveloped analytical model based on unstructured data, machine learning, and thefocused use of expert knowledge. DREAM overcomes the disadvantages of thetraditional approach when creating a causes of delay list, and enables thediscovering of the root causes of delay by applying a unique feature - temporaldistribution of the causes of delay. In a practical sense, the proposed modelprovides unbiased support in reconstructing the events related to delay at thesingle project level and its entities, which contributes to the reduction of disputesbetween contracting parties and aide intelligent decision-making on futureprojects. Грађевинарство - Менаџмент, технологије и управљање пројектима у грађевинарству / Civil Engineering - Management, Technologies and Project Management in Construction Datum odbrane: 07.07.2023. ашњење, базни узроци кашњења, изградња путне инфраструктуре, машинско учење, рударење по текстуалним документима, Transformer, неструктурирани подаци delay, root cause of delay, construction of road infrastructure, machine learning, text mining, Transformer, unstructured data 91552100 123419657 91552101 9396 2023-12-07T12:21:51.053Z 45 no 46 Marija, 1990- Ivanović 62851849 2023 63 mentor Zoran, 1966- Stojadinović 52055817 2023 63 član komisije Nenad, 1960- Ivanišević 12545127 2023 63 član komisije Milan, 1960- Trivunić 13372263 2023 63 član komisije Dejan, 1965- Marinković 52043785 2023 63 član komisije Đorđe, 1984- Nedeljković 52050697 2023 XVI, 163 стр. 3261719 http://phaidrabg.bg.ac.rs/o:31954 no yes 12 2023-12-07T12:21:51.320Z 70 11 1067197 1067293 1067313 кашњење, базни узроци кашњења, изградња путнеинфраструктуре, машинско учење, рударење по текстуалним документима,Transformer, неструктурирани подаци delay, root cause of delay, construction of road infrastructure, machinelearning, text mining, Transformer, unstructured data 624+69:005.8(043.3) 1738 11A18 2023