Preliminary Qualitative Analysis and Implications of Wood Products Perception on Social Media Preliminarna

• The article presents the results of the qualitative research of social media, managed by the Institute of the Civil Society, University of Ss. Cyril and Methodius in Trnava, in cooperation with the Slovak University of Technology in Bratislava. The research aimed to analyse different areas of the current management challenges and their perception of the selected social networks. The study concentrates on the presentation of the chosen manufacturers of the automotive industry and furniture industry on social media. The content analysis was based on the VADER (Valence Aware Dictionary and Entiment Reasoner) lexicon that was explicitly tuned to sentiments expressed in social media and QDA software.


UVOD
Social networking through online media can be understood as a variety of digital sources of information that are created, initiated, circulated, and consumed by Internet users as a way to educate one another about products, brands, services, personalities and issues (Chauhan and Pillai, 2013). As stated in the research of Enginkaya and Yılmaz (2014), social media integrates consumers with their own voice, not as passive respondents in their relationships with brands as in the past, rather as active members of brand communities (Miller and Lammas 2010). These attributes enable brands to reach the right people, in the right place and at the right time (Schivinski and Dabrowski, 2016). According to Karlíček and Král (2011) cited in Čeněk et al. (2016), the most notable positive qualities of online marketing are the possibility of precise consumer targeting, personalisation, interactivity, multimedia content, simple effi ciency measurement, and relatively low costs. Marketing the brands through social media is becoming precise, personal, interesting, interactive and social (Sri et al., 2011). During the last fi ve years, social media in marketing has become a subject of many studies and interests (see Figure 1 representing the worldwide search interest on Google).
This study demonstrates partial results from the research conducted by the Institute of Civil Society, University of Ss. Cyril and Methodius in Trnava. The research is oriented to the qualitative analysis of social media and their utilisation by different industries. Firstly the qualitative analysis was applied for the automotive industry (Slovakia is the leading car producer per 1000 habitants in the world). The results of the qualitative analysis have been published in Babčanová et al. (2019) and Šujanová et al. (2019). This qualitative analysis has been extended to the comparative analysis between the automotive industry and selected producers of the consumer durables like furniture or electronics.
Presented results of the text content analysis of the selected social media for the furniture industry are based on one-week data collection (tweets) and customer comments published on YouTube for selected manufacturers. The content analysis was based on the VADER (Valence Aware Dictionary and Entiment Reasoner) lexicon that was explicitly attuned to sentiments expressed in social media (Hutto and Gilbert, 2014) and QDA software (MAXQDA).
This paper is divided into the following sections: The fi rst part presents a literature review supporting the conceptual framework of the study. The second part is dedicated to the description of the research methodology as well as applied tools and data sources. The third part contains the results of the quantitative analysis of the social media content and comparison of the two selected industries. The fi nal section provides the author's conclusion about the research results.

MATERIJALI I METODE
The goal of the research was to analyse the content of the selected social media and compare the results for two different industries. Data collection was made using:     (Bekker, 2019). From the top best-selling car manufacturers in Europe in 2018, we have randomly selected fi ve manufacturers. The same principle of randomness was applied for the selection of furniture manufacturers in Europe where the list of fi rms, analysed in the Centre for Industrial Studies report (CSIL, 2019), was used.

REZULTATI
To use the offi cial channels of YouTube and Twitter media, authors have created a list of representation of the selected manufacturers on social media (see Table 1). The list was based on the information obtained from the offi cial web sites of the companies (Calligaris, Nobia, HSL, Howdens, Molteni&C, Lexus, Renault, SEAT, Peugeot, BMW).
As we can see, Facebook, YouTube and Instagram are the most used social media, while LinkedIn and Pinterest are not so popular.

YouTube
For the content analysis of YouTube, we have used offi cial channels obtained from the manufacturer's official web sites. Each channel was analysed according to the number of subscribers and videos (see Table 2).
There is a signifi cant difference between the car and furniture manufacturers, whereas the number of subscribers for car manufacturers varies between 24 400 and 975 000, it is from 9 to 1520 for furniture manufacturers. The data used for the analysis was obtained from the most popular video of the channel. Again, authors analyse data according to the number of views, likes, dislikes and comments for each video (see Table 3).
Comparing the absolute values (number of views) of the car and furniture manufacturers' most popular video, we can conclude that there is again a signifi cant difference. A closer look on the percentage of those that like or dislike the video gives us a different result, whereas for Howden the number of views was 451, around 69 % gave the "Like to this video". The highest percentage of "Likes" for the car producers was given to BMW, and it was just 0.37 %.

Twitter
The use of QDA software limited twitter analysis of one batch to: -Maximum 10000 tweets -Maximum 7 day period.   Tweets represents the number of tweets published by the manufacturer on the manufacturer offi cial Tweeter account during the monitored period. / Broj tvitova obuhvaća broj onih koje je proizvođač objavio na službenom Tweeter-računu tijekom praćenog razdoblja.
As VADER lexicon was used, another limitation was related to tweets in English.
We have collected two data sets: - Using the name of the manufacturer as a keyword (see Table 4) - Using the name of the Twitter account (see Table 5).
The data was collected from February 9th 2019 to August 8th 2019.
As well as for YouTube, Twitter results are very different between the industries. Twitter activities on the car manufacturer's offi cial accounts are hardly comparable to the activities of the furniture manufacturers.
We can see that the furniture manufacturer does not use Twitter as a channel for sharing the information with the customers.

Tweeterova analiza dojmova s leksikonskim pristupom
For the lexicon-based sentiment analysis, the text was extracted from tweets. After removing the tweets ..... Šujanová, Nováková, Pavlendová, Cagáňová, Canet:Preliminary Qualitative Analysis... that did not fulfi l the criteria (keyword or address) and freeing the text of non-applicable graphics, we counted word frequencies in tweets using the Go list dictionary function of the MAXQDA software. VADER lexicon was used as a dictionary. The same operation, word count, was also applied without the VADER lexicon. The results of the word frequencies are presented in Table 6.
For car manufacturers, there is a signifi cant difference between the SEAT and other manufacturers. The tweets on the SEAT offi cial account contained signifi cantly more sentiment words than those of the rest of analysed car manufacturers.

Lexus
Lexus tweets contained 20 words from the VAD-ER lexicon, where 19 had a positive average value, and one word (limited) had a negative average value.
The fi ve words and emoticons with the highest average positive value were: best, greatest, love, amazing, win.

Renault
Renault tweets contained 73 words from the VADER lexicon, where 68 had a positive average value and 5 words (demand, limited, no, accidents, mistakes) had a negative average value.
The fi ve words and emoticons with the highest average positive value were: best, love, excellence, great, amazing.

SEAT
SEAT tweets contained 15 words from the VAD-ER lexicon, all of them with the positive average value.
The fi ve words and emoticons with the highest average positive value were: best, :*, kind, care and :).

Peugeot
Peugeot tweets contained 13 words from the VADER lexicon, where 12 had a positive average value, and one word (broken) had a negative average value.
The fi ve words and emoticons with the highest average positive value were: exciting, glad, *:, easy, sparkle.

BMW
BMW tweets contained 48 words from the VAD-ER lexicon, where 43 had a positive average value and 5 words (cutting, limited, no, diffi cult, problems) had a negative average value.
The fi ve words and emoticons with the highest average positive value were best, excellence, great, happy, pleasure.

ZAKLJUČAK
From the very beginning, we have to acknowledge that to compare two very different industries (automotive and furniture) applying just quantitative data brought the expected result: automotive industry representation on social media is incomparable with the representation of furniture industry. Where, for example, the number of subscribers to the offi cial channels on YouTube for the automotive industry ranges between tens of thousands and hundreds of thousands, in the case of furniture industry the number ranges from tens to thousands of subscribers.
Differences can also be observed in the range of supported social media. For the automotive industry, all manufacturers have a Twitter account, for the furniture industry, just three of those selected. A different situation is seen with Pinterest, where most of the furniture producers have an account, while among the car producers, just Lexus has such an account.
Twitter analysis brought similar results. Selected representatives of furniture manufacturers have been mentioned in the tweets during the analysed period from 0 to 37 times. In contrast, in automotive industry, this number ranged from 4381 to more than ten thousand.
Differences have also been observed in activities on the offi cial Twitter accounts, where there was no activity during the analysed period for furniture manufacturers. Therefore, there is no comparison between the content of the tweets applying the sentiment analysis with the lexicon-based approach. It can just be concluded that, for the automotive industry, the content of the tweets on the offi cial tweet accounts contained mostly positive words from the VADER lexicon. The word with the highest positive average value was best (four manufacturers), followed by excellence, kind, and care (two manufacturers). Authors have not expected such signifi cant differences in the use of social media between the selected car and furniture manufacturers considering the fact that selected manufacturers are listed between the top best-selling producers for the year 2018. It is evident that the automotive industry is widely exploiting social media and customers' opinion data for the innovation aimed to fulfi l customers' needs and expectations. As for the furniture industry, the potential of the social media infl uence on consumer behaviour, according to the preliminary research results, was not suffi ciently recognised.
The present research will continue with the detailed sentiment analysis using Twitter and YouTube data for a larger group of manufacturers during at least ten discrete periods.