Search

Personality adjectives are important in trait theories, but their significance in everyday language use is understudied. Hence, we compiled a set of 706 English adjectives previously linked to Big Five traits and analysed their frequency of occurrence in social media using the dataset PANDORA, which contains 14 million sentences from Reddit. The analysis is funnelled and proceeds in three stages, i.e., we count the occurrence of 1) the personality adjective alone (e.g., kind) ; 2) bigrams of adjectives and the noun "person" or its equivalent substitutions (e.g., kind dude) ; 3) the adjectives in self-descriptive sentences (e.g., I'm a kind dude.). Almost 1/4 of the adjectives were unmentioned and frequencies of occurrences were uneven between traits and between their positive and negative poles. We take a closer look at the least (e.g., uninquisitive, self-pitying) and most frequently used adjectives (e.g., simple, smart) and discuss the opportunities and challenges of analysing online language.

Recently, advances in natural language processing (NLP) and machine learning have enabled research on personality cues that are ubiquitous in social media texts. In this study, we investigate whether the relationship between religiousness and personality traits demonstrated in the literature can also be found in social media texts using NLP. We propose a method based on the implementation of the statement-to-item matching framework (Gjurkovic et al., 2022), which we use to filter over 14 million sentences written by users with self-reported Big5 scores from the PANDORA Reddit dataset to find the most informative self-disclosures of users about their religious affiliation. Based on the filtered sentences alone, raters were able to classify a subset of users (n=299) into religious and non-religious groups. Although median differences on personality traits were in the expected direction, these differences were not significant. We present implications, challenges, and opportunities for using this method in future studies.

Automatska procjena ličnosti na temelju teksta s društvenih medija dobiva sve veću pozornost kako u psihologiji tako i u području umjetne inteligencije. S jedne strane, interes psihologa primarno pobuđuje mogućnost korištenja digitalnih ponašajnih tragova za procjenu ličnosti. S druge strane, zbog brzog rasta količine tekstnih podataka koje generiraju korisnici interneta, stručnjacima računarske znanosti se kao zanimljiv izazov nameće rad s takvim velikim i nestrukturiranim podacima. No, trenutačni automatski pristupi procjenjivanju ličnosti nisu usmjereni k osiguravanju interpretabilnosti (informacija koji su točno podaci relevantni za procjenu ličnosti) i valjanosti (informacija jesu li korišteni podaci zaista valjani tragovi ličnosti). Upitnici ličnosti, međutim, kao nužni temelj imaju upravo interpretabilnost i valjanost. Kako bismo ublažili navedene slabosti automatske procjene ličnosti, predlažemo pristup koji kombinira upitničku i automatsku procjenu ličnosti. Naš pristup akronima SIMPA (engl. Statement-to-Item Matching Personality Assessment) koristi metode obrade prirodnoga jezika kako bi detektirao samoopise ličnosti koji se potom koriste za automatsku procjen ličnosti. Srž pristupa jest pojam dispozicijske semantičke sličnosti između slobodno izraženih izjava i upitničkih čestica. Takva sličnost kombinira semantičku sličnost sa znanjem o načinu na koji bi se određena dispozicija mogla manifestirati. Konceptualnu osnovu pristupa SIMPA čini model realistične točnosti (Funder, 1995), koji opisuje korake u procesu dolaska do točne procjene ličnosti, a kojeg proširujemo mehanizmom povratne petlje koja dodatno poboljšava točnost procjene. U izlaganju predstavljamo jednostavnu implementaciju pristupa SIMPA na podacima s društvene mreže Reddit. Demonstriramo kako se pristup može koristiti izravno za procjenu velikih pet crta ličnosti kod korisnika Reddita, što potvrđuju statistički značajne korelacije između tako procijenjenih crta i samoprocjena istih crta. Pristup također koristimo neizravno za proizvodnju značajki za nadzirani model strojnog učenja za automatsku procjenu ličnosti, čime dobivamo trenutačno najbolje rezultate na zadatku predviđanje ličnosti korisnika Reddita. Konačno, raspravljamo o važnim mogućnostima i izazovima analize jezika na internetu.

Computational methods have found a wide area of application in social sciences and humanities. These methods focus on collecting and analysing large quantities of data gleaned from digital environments. Although such methods are interdisciplinary and diverse, they all tend to share three common features: automated or semi-automated collection of data from digital environments, large scale datasets, and the usage of a range of different techniques drawn from computer science to model and analyse the data. Yet, beyond the argued impact in re-defining scientific method (Anderson, 2008 ; Floridi, 2014 ; Kitchin, 2014), problems around the applicability and epistemology of computational methods were soon outlined by some scholars (boyd and Crawford, 2012). A lack of theoretically informed inferences (Coveney, Dougherty, Highfield, 2016 ; Boyd, 2018) is often found in some studies, supported by an assumption that computational methods allow data to speak for itself. Recent scholarship has found a fruitful connection between theoretical elements of the political economy of communication (PEC) and various computational techniques (e.g. Bilić, Furman, Yildirim, 2018 ; Furman, Saka, Yildirim, Elbeyi, 2019 ; Birkinbine and Gomez, 2020). PEC provides a broad conceptualisation of the role of the media in society. It looks at the structural position (i.e. media ownership) and connects it with opportunities, constraints and limits to publishing and framing news content in certain ways. In this presentation, we use PEC to select media outlets based on audience reach, regional coverage, public ownership, and non-profit ownership with media frame analysis that looks at the salience of specific linguistic elements in news reports. This allows us to make theoretically-informed inferences on the thematic scope of news media output in relation to the COVID-19 pandemic. Comparing two peak periods of the COVID-19 pandemic during 2020, the empirical part of our study will apply a computational methodology and prepare a semantic textual analysis of all COVID-19 related articles published by 21 Croatian digital news media outlets. This methodology will use natural language processing and correspondence analysis to explore which news concepts relate to which media, and which news themes and frames dominated two peak periods in relation to different structural positions of the media. The findings of our exploratory study will be presented through visual representations of the complex dataset.

Potrošnja betona je raširena u svim zemljama svijeta, pa je kao takav jedan od najvažnijih građevinskih materijala. Uvođenjem ojačanja, značajno je povećanje vlačne i savojne čvrstoće, što krhkom materijalu kao što je beton omogućava preuzimanje većih opterećenja. Od davnina su poznata mikro ojačanja u obliku vlakana, a u novije vrijeme, uz upotrebu čeličnih i sintetičnih vlakana, se javlja zahtjev za ekološki prihvatljivijim - prirodnim vlaknima. U ovom radu ispitati će se svojstva morta i betona ojačanih vlaknima brnistre, te se dobiveni rezultati usporediti sa njihovim standardnim mješavinama., Concrete consumption is widespread in all countries of the world, and as such is one of the most important building materials. By introducing reinforcement, it is significant to increase the tensile and flexural strength, which allows a more fragile material, such as concrete, to withstand higher loads. Micro fiber reinforcements have been known since ancient times, and more recently, with the use of steel and synthetic fibers, there has been a demand for more environmentally friendly - natural fibers. In this paper, the properties of mortar and concrete reinforced with fibers of the broom will be examined, and the results obtained will be compared with their standard mixtures.

Cilj ovog istraživanja bio je provjeriti ulogu međuspolnog odabira u iskazivanju altruističnog ponašanja. Prethodna istraživanja pokazala su da su ljudi altruistični prema onima s visokom vrijednosti kao partnera, iz čega se zaključuje da altruistično ponašanje može služiti u svrhu privlačenja partnera. No dosadašnja istraživanja su zanemarila važne varijable kao što su strategija izbora partnera i socio-ekonomski status (SES) osobe kojoj je potrebna pomoć. Ovo istraživanje usmjereno je na žensko altruistično ponašanje prema muškarcima, koje je slabije istraživano od altruističnog ponašanja muškaraca prema ženama. Istraživanje je provedeno online anketom, a uzorak su činile 1522 sudionice. Zadatak sudionica bio je procijeniti namjeru altruističnog ponašanja prema muškarcima na fotografijama koji su se razlikovali s obzirom na tjelesnu privlačnost i SES (operacionaliziran odjećom). Rezultati su pokazali da žene imaju veću namjeru altruističnog ponašanja prema muškarcima visoke nego niske tjelesne privlačnosti te prema muškarcima visokog nego niskog SES-a. Efekt veće namjere altruističnog ponašanja prema visoko tjelesno privlačnim muškarcima, u odnosu na one niske tjelesne privlačnosti bio je izraženiji za sudionice s kratkoročnom strategijom izbora partnera nego za one s dugoročnom. Dobivena je zanimljiva interakcija tjelesne privlačnosti i SES-a muškaraca, a interakcija strategije izbora partnera sudionica i SES-a muškaraca je izostala. Općenito je ustanovljena značajna uloga međuspolnog odabira u altruističnom ponašanju. The aim of this study was to investigate the role of inter-sexual selection in altruistic behaviour. The results of previous research show that individuals are more altruistic towards others with high mate value, suggesting that altruistic behaviour may serve as a mating signal. However, these studies neglected important variables such as mating strategy and socioeconomic status (SES) of a person in need. This study examined exclusively women’s altruistic behaviour towards men, since it is explored to a lesser extent than men’s altruistic behaviour towards women. The research was conducted online and the sample size was 1522. Participants were asked to rate the likelihood of altruistic behaviour towards men on photographs, who differed in physical attractiveness and SES (manipulated by clothing). The results show that women generally had a higher intention to help the physically attractive than unattractive men, but this effect was more pronounced for women with a short-term mating strategy. Also, men with high SES got higher scores on the altruistic behaviour intention than those with low SES. Physical attractiveness and SES yielded an interesting significant interaction, while the interaction between SES and mate strategy was insignificant. Generally, the results support the role of inter-sexual selection in altruistic behaviour.

U radu je prikazana analiza protoka rijeke Vrljike mjerenih na stanici Opačac. Na temelju mjerenih dnevnih podataka u razdoblju od 1995.-2000. godine dani su grafički prikazi histograma otjecanja, krivulje trajanja protoka, transformirane krivulje trajanje protoka te sumarni prikaz krivulje trajanja i sumarni prikaz transformirane krivulje trajanja protoka za navedeno razdoblje., This study presents the hydrological analysis of river Vrljika (the karst spring Opačac). Graphical representations of hydrograph are given based on the measured daily data in period 1995-2000 year, and it was performed statistical data processing and analysis of the histogram, flow duration curves, transformed flow duration curves, summary display of flow duration curve and summary display of transformed display curve for given period.