Over the years, the World Wide Web has emerged as the primary means of communication and information sharing, resulting in significant shifts in how people understand the process of consuming information [1] . In this contemporary landscape, information about artistic works and cultural heritage artifacts has also migrated into the digital realm through digitization [2] . To provide users with an engaging experience when interacting with content related to art and culture, it is crucial to comprehend their behavioral patterns [3] and to utilize that insight in crafting a user experience (UX) that caters to the distinct needs of both the users [4] and the content itself. This necessity is particularly heightened in outlets presenting vast amounts of information, as the user interface (UI) is an essential instrument to mitigate user confusion [5] and enhance engagement by prioritizing simplicity and familiarity.

It is safe to say that a key aspect of sharing cultural events, including various audiovisual art projects, is their visibility on the Internet [6] . Online media platforms, cultural websites, blogs, social media, and other digital repositories offer abundant information about art events. However, news coverage available on the web is frequently both scattered and short-lived, influenced by the differing interests of traditional media and user-generated content [7] [8] , as well as the popularity of algorithm-driven news distribution systems [9] . To gather more meaningful insights into the traits of modern audiovisual art projects, it is essential to extract information from a variety of sources. Additionally, the process of pinpointing and gathering details regarding a specific event from different websites can result in a well-informed assessment of its potential reach, which may subsequently affect its cultural significance [10] .

In order to collect large-scale information from the Web in an automated manner, it is essential to programmatically analyze the structure of Web pages and their metadata, as they appear in social media data-graphs or Semantic Web integrations [11] . Additionally, this data must be recorded in a manner that takes advantage of the benefits of Semantic Web ontologies and data models. The Semantic Web arose as an effort to enhance how information is structured online to improve its ability to convey meanings in a manner that digital systems can interpret. This process includes the systematic arrangement of data into standardized, subject-oriented hierarchies known as ontologies [12] . Ontologies serve as a means of comprehending and organizing web information and lie at the heart of the Semantic Web, facilitating the development of a versatile digital space that acts not just as a collection of relevant data but as a dynamic, intelligent, and efficient environment with new opportunities for advanced applications and services [13] . Although computers do not “understand” information like humans do, they can efficiently relate and manage it so that the outcome becomes meaningful for human users [14] . Scholars in Web science are continuously working on research and technological advancements to actualize the Semantic Web concept [15] . For instance, the RDF/XML standard has shown to be challenging to apply in scenarios with high complexity [16] , whereas newer technologies such as N-Triplets [17] , Turtle [18] , and JSON-LD [19] offer more readable structures, enabling better grammar and smoother mapping from XML syntax to RDF. As a result, the technologies capable of realizing the essential Semantic Web are evolving and will continue to develop, moving closer to achieving semantic understanding of data by machines.

Going one step further, Generative Artificial Intelligence (GenAI) can be used to both analyze html documents and derive structured information from their texts. AI technologies have a long history of successful application on text structuring and summarization [20] which has become even more powerful with the advances in AI and the introduction of the transformer architecture, which gave birth to GenAI as we know it [21] .

Mining information from the Web may lead to large objective datasets which are quintessential in statistical analysis. In recent years, there has been a growing interest among researchers around big data and computational engineering. Big data is defined as an extensive collection of datasets that conventional data processing systems cannot analyze; instead, they demand innovative automated technologies for tasks like capturing, storing, distributing, managing, and analyzing [22] . Online news items are an example of such datasets. This vast and varied information is unmanageable through manual processing, which involves breaking it down into sentences or words to analyze and extract valuable insights [23] . Additionally, significant attention from researchers has been directed toward the application of big data, highlighting its role as a crucial indicator of a company’s efficiency and productivity [24] . This use of big data has also found its way into social sciences and culture in the form of cultural analytics which may be defined as “the analysis of massive cultural data sets and flows using computational and visualization techniques” [25] .

In this study, modern innovative technologies are utilized to create a large objective dataset from the Web concerning audiovisual art events as they are presented by online media outlets and other related websites. Via the use of programmatic Web Search, Semantic Web analysis and the power of GenAI structured information is collected about the various aspects of an event’s online presence. This information is quantified and processed to derive objective measurements, which are used for statistical analysis. This article presents the detailed methodology of both data collection and an analysis of real-world data regarding contemporary works of audio and visual art in Greece. The collected data are openly provided to the general public through an innovative Web application which focuses on user experience and functionality [26] available at https://repo.artdata.gr/ and through an extended application programming interface. The analysis focuses on the classification of the collected events based on the event type, the methods and techniques used in the event as well as the events’ thematic elements. By investigating both the popularity of specific classification tags and the popularity and perceived cultural outreach of individual events described by these tags, the study demonstrates how objective Web data may be collected, analyzed and used to reach significant conclusions about the importance of various artistic subjects, methods and media in contemporary culture. The following research questions concerning the Greek audiovisual event landscape were established as a means to guide the research direction:

RQ1. What are the most popular classifications of audiovisual art events regarding event type, methods, and subjects?

RQ2. What are the classifications of audiovisual art events regarding event type, methods and subjects that are most influential based on event outreach index?

The algorithm for data extraction is made up of three primary components. The initial phase involves searching through media websites, which, after being processed, results in a collection of audiovisual events. The second and most important component includes an iterative process that seeks additional publications related to each of these events on the internet, transforming the content of these publications from code and free text found on any webpage into a graph of structured data that is integrated into the recorded information about the event in the system’s database. Additionally, during this phase, publications that are extracted but do not relate to an event contribute to the formation of new events for further investigation. Lastly, the third component investigates the websites where the publications retrieved during the event inquiry originated and gathers information from them.

During a data collection period that took place in two separate phases in the Spring and Summer of 2024 over 16,500 different audiovisual art events in Greece were discovered. These events were reported in over 30 k unique webpages from over 2.4 k different media outlets and websites. The generation of classification tags based on the texts of these discovered publications lead to over 1000 different tags describing the type of the events, over 1700 different tags describing the methods and techniques used in the events and over 4500 different tags describing the events’ themes and subject matter. Table 1 presents a summary of the collected information.

In order to gain insights into the popularity of the various classification tags the total sum of events which were described using one such tag was calculated. This led to a ranking based on how plentiful were the events which belonged to that specific tag. Table 2 presents the top 20 most popular tags describing an event’s type. Since the data extraction process involved Greek events and publications mainly in the Greek language, the tag names have been translated into English for the reader’s benefit.

The distribution of events over these tags is presented more clearly in Figure 2 in the form of a pie chart. According to the chart music related events are dominant in the landscape of audiovisual art events, with the first most popular tags “music” and “concert” culminating in over 40% of the events. The performative arts follow with “theater” and “cinema” being in the top 5 most popular tags alongside the more generic tag of “culture”. The top 20 is populated with other general types of art events such as “exhibitions”, “visual art” and “dance”, but also includes more specific subcategories such as “comedy”, “classical music” and “drama”.

Going beyond simple event type categorization, the top 20 tags describing the means, methods and techniques used during the various audiovisual art events are presented in Table 3 .

How the various tags relating to audiovisual art methods are distributed over the discovered events is presented more clearly in Figure 3 in the form of a pie chart.

Continuing the trend observed in Figure 2 , the more musical methods are dominant with “live music” getting first place with over 20% of the events and accompanied by other related tags such as “music”, “concert”, “vocals” and more. The performative art methods and means also play a dominant role with “theatrical performance”, “direction” and “acting” all appearing in the top 5. The chart also includes the performance of various musical instruments such as “piano”, “guitar” and “drums”, as well as methods of storytelling, including “dance”, “dramaturgy”, “narration” and “speech”. “Painting” is also present in the top 20 as a representative of the fine arts.

Finally, the third category of classification tags included the themes and subjects of the events. Table 4 presents the top 20 most popular subject tags and the number of events described by each tag.

The distribution of the above tags can be presented in a more comprehensive manner through the pie chart available in Figure 4 .

As presented in Figure 4 the importance of musical events is conveyed by the multitude of musical styles that are dominant in the top 20 subject tags. Going beyond the very general “music” and “concert” we get “Greek music”, “musical tradition”, “classical music” and “Laïko music” (which is a very prevalent blend of folk and popular music in Greece) [27] . “Culture” and “cultural heritage” are also popular tags and alongside “history” and “tradition” indicate a tendency for many events to focus thematically on Greece’s rich cultural past. Themes and subjects popular in performative art are also included in the chart with “social issues”, “love”, “romance”, “friendship” and “personal relationships” all being prevalent in the top 20 most popular subject tags.

Tables 2-4 and Figures 2-4 give an interesting glimpse on the landscape of audiovisual arts in contemporary Greece, based on the popularity of the various classification tags that were mined from texts reporting on audiovisual art events on the Web. These real-world data represent an objective reality as reported through the web. Music, performative art, and other forms of expression are all represented in various measurable degrees. But the number of events for each specific classification tag cannot be the only indicator. The difference in popularity, outreach and cultural impact between various events also plays a significant role.

Out of the total 16,501 events a total of 2685 were fully investigated in terms of outreach using the web search algorithm to collect exhaustive information about them as described in the methodology section and using the outreach analysis algorithm to calculate their various outreach metrics. The culmination of these outreach measurements in one singular index allowed us to gauge the popularity of each event. For each classification tag that appeared in at least ten different fully investigated events an average outreach index was calculated. The threshold of ten events was established in order to have a representative sample of events before calculating the average. Figure 5 presents the top 20 more impactful event type as a bar chart of the average outreach index of each type.

Through the bar chart of the top 20 type tags with the highest average outreach index a series of highly impactful event type subcategories are established. Musical event genres such as “pop”, “rock”, “classical”, “opera” and “Greek music” all get higher places than the wider general tags like “music” got, indicating a higher outreach of these specific genres. Additionally, “audiovisual art” as a tag makes a strong appearance indicating a higher outreach for events with a clearer artistic identity. At the same time niche types like “improvisation” or “animation” make an appearance, and this is rounded up by some more general themes like “charity” or “LGBTQ+” issue related events.

Moving on to the audiovisual event classification tags for methods, techniques and media, Figure 6 presents the top 20 tags with the highest average outreach index as a bar chart.

As made clear by the bar chart, more technical presentation methods and techniques utilizing “audiovisual effects”, “visual effects”, “audiovisual media” and “electronic instruments” present a high average outreach index. Moreover, general technical aspects of an event are also highly important with “sound effects”, “stage presentation”, “musical arrangement” and “instrumental performance” making an appearance in the top spots. The high production values related to the use of modern technology and high technical implementation appear to have a big impact on the outreach of the events.

Finally, through Figure 7 , we take a glimpse into the top 20 event classification tags for subjects and themes with the highest average outreach index in the form of a bar chart.

On the matter of subject classification tags, seasonal or annual themes have the highest average outreach index. The annual “Eurovision” song contest with its multitude of preliminary, semifinal, and final events appears in the top spot, followed by “New Year’s”, “celebration”, “anniversary” and “anniversary concert” in the top 20, indicating the high outreach of seasonal events. The tag “Maria Callas” which also appears on the top 20 list is related to a multitude of events in tribute to 100 years from the artist’s birth, which also adds to the importance of anniversary related events. Music themes revolving around the “artist collaboration”, the “artistic collaboration” and “music career” point to a high outreach for events that have added significance due to scope or multitude of participants, while popular musical genres such as “pop”, “pop music” and “Greek songs” also allow for high outreach. Some tags related to “Greek culture”, such as “Greek tradition”, and “musical heritage” also make an appearance.

The tables and figures presented in the results section clearly show that the popularity of the classification tag itself and the popularity of the events that are described by such a tag in terms of outreach on the World Wide Web are significantly different and as such they present a challenge in terms of understanding the importance of each tag. This will be further discussed in this section.

This study focuses on the use of contemporary real-world data, extracted from the Web through an advanced algorithm utilizing semantic web, search engine and generative AI technologies, for the purpose of classifying audiovisual art events. Using objective quantitative metrics, the event data were distilled into a singular web outreach index, which quantified their perceived impact based on their online presence. Using the events classification in terms of event type, methods and techniques, and thematic subjects, the popularity of various classification tags was explored. Going one step further, the impact of the events described by each tag was estimated using each event’s outreach index. This process was used to discover important classification tags in terms of potential influence.

The analysis of the popularity of various classification tags within the current context of audiovisual art events in Greece, using a data-driven quantitative approach, objectively validated the significance of music and performative arts, alongside the popularity of particular musical genres and narrative themes focused on interpersonal relationships. Furthermore, examining the average outreach index for each classification tag highlighted the relevance of Greek musical heritage, the benefits of utilizing modern technology in performances, and the additional value associated with events that are annual, seasonal, or commemorative in nature. In addition to the insights discussed, the findings of the study encompass the methodology used, as well as the dataset collected, both of which can be applied similarly to yield data-driven discoveries to other research questions with different goals.

The focus of both the data extraction process and the classification and outreach analysis is the various events’ online presence. The information available on the Web allows an objective numerical representation of the events but offers limited insight about events that may be important and culturally impactful, but disproportionately promoted online. This limitation of the study is somewhat mitigated by the extraction of information from user generated content, blogs, reviews etc., but remains important. For this reason, it is optimal to use data-driven quantitative analysis in tandem with qualitative expert analysis to combine findings and reach safer conclusions.

In the future, the collected dataset may be used in addressing issues beyond classification, looking into geographical or temporal attributes of audiovisual art events. Moreover, the analysis may move from a descriptive statistical discussion to the use of computational statistics and machine learning in order to gain deeper insight. In the next phase of this research, techniques such as clustering and topic modeling will be used to gain better understanding of the dataset, based both on its quantitative characteristics and its descriptive texts.

Obtaining a clearer understanding of the landscape of audiovisual art events using real-world quantitative metrics, in tandem with a critical qualitative approach of the wider field of art and culture, can provide the basis for more informed decision making from, the artists, the organizers and the audience. This clearer approach may act for the benefit of everyone involved, leading to more appealing, interesting, and lucrative events and to a deeper relationship between the artistic product and the general public.

This research was funded by the Hellenic Foundation for Research & Innovation, within the framework of the “2nd Call for H.F.R.I.’s Research Projects to Support Faculty Members & Researchers” (project number 3607).