Browsing by Issue Date, starting with "2025-03-03"
Now showing 1 - 2 of 2
Results Per Page
Sort Options
- Effects of road density on regional food websPublication . Mestre, Frederico; Bastazini, V. A. G.; Ascensão, F.Roads stand as major threats to biodiversity because they affect the functioning of ecosystems and the provision of ecosystem services. Understanding how the effects of roads affect the dynamics of ecological interactions is essential to managing human impacts on biodiversity, but such studies are few. We investigated species vulnerability to road density and effects of road density on food webs across Europe. Using species-specific road density thresholds, beyond which local populations may not persist, and trophic interaction data (predator-prey interactions), we constructed regional food webs to assess the potential loss of trophic interactions due to roadkill. We analyzed data on 551 species across top, intermediate, and basal trophic levels. Effects of roads varied spatially. In areas near major cities, species lost >90% of their trophic interactions. We found 191 species that were affected by loss of prey or predators. Apex predators exhibited lower direct impacts from road density than predators at lower trophic levels, and basal-level species seemed more exposed to direct road-related effects (roadkill), which could trigger a cascade of interaction disruptions. Our findings emphasize the need for informed road infrastructure development and targeted conservation strategies to mitigate the negative impacts of roads and traffic and thereby preserve the integrity of ecological networks. Our identification of critical areas where road-induced cascade effects may be most pronounced and of groups of species that may be at higher risk from roads can inform policy and conservation planning.
- Engagement models to monitor brand activationPublication . Lemos, Marco Matias de; Cardoso, Pedro J. S.; Rodrigues, João M. F.Engagement can refer to the act of being involved or committed to a particular activity, organization, or relationship. It also relates to the involvement and enthusiasm of em ployees or customers to a brand or to an event. The creation of effective brand names and the assessment of audience engagement are critical for businesses aiming to estab lish unique market identities and foster consumer connections. This dissertation ex plores (i) the transformative potential of artificial intelligence (AI), particularly natural language processing (NLP) and affective computing (AffC), in innovating brand name generation and engagement evaluation processes. Addressing the challenges posed by traditional approaches, such as trademark constraints and the subjective evaluation of branding elements, an AI-driven model is introduced to quantify the "brandabil ity" of two-word combinations. By leveraging psycholinguistic dimensions – valence, arousal, and dominance – alongside linguistic and semantic attributes like concrete ness, word frequency, and cosine similarity, the model evaluates the impact of over 219.000 word pairs, offering robust predictive capabilities for data-driven branding. Also, focusing in image/video this dissertation explores (ii) the engagement in group events, i.e., proposes a group engagement model that integrates individual metrics such as gaze direction, valence, and arousal to classify engagement states into binary levels – engaged and not-engaged – with further sub-level distinctions. Experimental results highlight the model’s capacity to assess and adapt to dynamic scenarios.