SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Related Research
  5. Data Collection and Analysis
  6. Results and Discussion
  7. Conclusions
  8. Acknowledgements
  9. References

Two sites for tagging, one in China (i.e., 365Key) and one in the USA (i.e., Del.icio.us) are compared in terms of tagging mechanisms and tags created. In general, the Chinese tagging site provides its users with more pre-set functions whereas its American counterpart gives more freedom to its taggers. Our findings also show that tagging, like many other information behaviors, is greatly influenced by and stamped with the social and cultural traditions existing in each country. Taggers in both countries, however, do tend to choose terms of same or similar meanings, indicating that tagging, regardless of where it is done and where the tagger is from, is usually done according to the fundamental rules in indexing (e.g., nouns or noun phrases as tags). On the other hand, tagging as an activity unique in the networked environment for loosely representing and organizing all kinds of information, does not seem equal to keyword indexing which has been done in producing database systems (e.g., InfoTrac) and search engines (e.g., Google).


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Related Research
  5. Data Collection and Analysis
  6. Results and Discussion
  7. Conclusions
  8. Acknowledgements
  9. References

Social tagging or tagging in brief, sometimes being referred to as social bookmarking, becomes more and more popular on the Internet. Tagging using a folksonomy, a term coined by Vander Wal (2004), represents a new approach to organizing net-based information. The Bulletin of the American Society for Information Science &Technology even published a special issue on folksonomies in October/November 2007. According to Rainie (2007), 28% online Americans have tagged on the Internet. In China, tagging is also becoming trendy among many Internet users. All these indicate that tagging is no longer just a technology. Rather, tagging becomes an emerging social phenomenon which attracts the attention of many Internet users.

In order to gain a better understanding of tagging in China and the USA, we selected two sites that support tagging, namely, http://del.icio.us and http://www.365key.com (Referred as 365Key hereafter), and intend to address the following research questions.

  • What features do these two tagging sites possess?

  • What tags were created at each site? And what can be gleaned from an analysis of those tags?

  • What similarities and differences can be found between these two sites for tagging?

  • What implications are there for tagging in China and the USA?

Related Research

  1. Top of page
  2. Abstract
  3. Introduction
  4. Related Research
  5. Data Collection and Analysis
  6. Results and Discussion
  7. Conclusions
  8. Acknowledgements
  9. References

Tagging is not new in the information age since people have been doing so in the past. For example, keyword indexing can be regarded as a synonym for tagging, and authors normally supply several keywords to “tag” the content of their manuscript submissions. Rather, it is tagging on a common, social platform (e.g., Del.icio.us and 365Key) by numerous people known as taggers that makes tagging becoming a noteworthy social phenomenon (Fitcher 2006). Golder and Huberman (2006) examined the structure and dynamics of collaborative tagging systems, and discovered regularities in users tagging activities (e.g., tag frequencies and kinds of tags created). Binowski (2006), on the other hand, found that social proof having an effect on tag selection in social bookmarking based on a study conducted at an academic institution. The focus of such studies seems on how to tag effectively with social bookmarking applications like Del.icio.us and Flickr.

With increasing participation of users in tagging activities, researchers also started examining other aspects of tagging such as analysis of tags and tagging behaviors. For instance, Kipp (2006; Kipp & Campbell 2006; 2007a; 2007b) did several studies on some popular social tagging sites (e.g., Del.icio.us, Citeulike). The findings of those studies include tagging behaviors, the role of non-subject tags in the tagging process, and the context of tagging from the perspectives of users, authors and intermediaries. Farooq and colleagues (2007) even proposed a metric for evaluating a tagging system. The authors also suggested some design heuristics for implementing a social bookmarking system. Muller (2007), targeting a different group, did a comparative study of four enterprise tag-based services.

As seen, little research has been done so far on tagging systems from different countries, letting alone to explore implications of such practices in different cultures. The current study is therefore carried out to examine two tagging sites, one in China and one in the USA, in order to find out the practices and implications of tagging in these two distinctive countries.

Data Collection and Analysis

  1. Top of page
  2. Abstract
  3. Introduction
  4. Related Research
  5. Data Collection and Analysis
  6. Results and Discussion
  7. Conclusions
  8. Acknowledgements
  9. References

Del.icio.us is one of the most popular social bookmarking sites in the USA while 365Key, being the first of its kind in China, enjoys a similar status in tagging as its USA counterpart - Del.icio.us. These two sites were thus chosen for the present study to explore and contrast their features, tags created at each site, and other aspects of tagging.

Python was used in writing a program for extracting tags from both sites in the News category. The News category was chosen because it is perhaps the most comprehensive as well as representative category at both sites given the fact that China and the USA are different from each other in many other dimensions. Tags considered in this study were extracted at both sites from 6:53pm on January 4, 2008 to 11:03am on January 5, 2008, by taking into account of the time difference in China and the USA. As a result, a total of 5064 tags were obtained from Del.icio.us and 13006 tags from 365Key. In consideration of feasibility, only tags with a usage frequency of 10 or more were selected and analyzed, which yields 89 tags for del.icio.us and 59 for 365Key.11

Tags collected from each site were then analyzed quantitatively and qualitatively to answer the research questions posed earlier. Besides, both tagging sites were carefully inspected as well from various perspectives to discern their functionalities and characteristics.

Results and Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Related Research
  5. Data Collection and Analysis
  6. Results and Discussion
  7. Conclusions
  8. Acknowledgements
  9. References

In this section, we will first compare the features of Del.icio.us and 365Key, followed by a content analysis of the tags chosen for this study. Similarities and differences between those two sites will be presented and discussed toward the end of this section.

Tagging Mechanisms

Both Del.icio.us and 365Key allow their users to create, organize, and browse tags. However, users at each site do tagging activities differently due to the site-specific supporting mechanism provided by either Del.icio.us or 365Key. Table 1 summarizes their similarities and differences in tagging.

Table 1. Tagging Features of Del.icio.us and 365Key
Thumbnail image of

As shown in Table 1, Del.icio.us and 365Keys provide users with different tagging mechanisms. For example, 365Key lets its users create more fields (e.g., title or URL) to place a tag than what Del.icio.us offers to its taggers. Meanwhile, users at 365Key can evaluate and comment on what they tag, thus recommending or not recommending what has been tagged. Tag organization is also implemented differently at these two sites. Although both can be said in adopting a general folksonomy (Vander Wal 2005), 365Key has additionally created some common categories for its users to choose from during tagging. Naturally, users can always tag without consulting the categories 365Key provides. However, tagging using a pre-built folksonomy might encourage tag reuse, which is one of the main factors considered in tagging evaluation (Farooq, et al. 2007).

Similarly, Del.icio.us users can tag at will or use tags already created by others. This approach awards tagging freedom to users but may cause some problems such as polysemy, synonyms and plurals in tagging (Noruzi 2006). In addition, Del.icio.us also allows its users to create tag bundles, a method for arranging previously-used tags into groups. This enables Del.icio.us users to manage their own tags using folders similar to folksonomy categories. As for other tag management functions, Del.icio.us users can rename and delete tags more easily than 365Key taggers. The latter have to first click certain tags before performing any such tasks.

On the other hand, 365Key taggers can browse tags by viewing or bookmarking frequencies at various time intervals (e.g., 12 hours, 24 hours, 3 days). One manager at 365Key (Hong 2005) explained at his blog, when referencing to the afore-depicted browsing features, that the viewing frequency represents the popularity of tagged contents whereas the bookmarking frequency indicates the value of tagged information.

Although both Del.icio.us and 365Key offer comparable mechanisms for their users to share tags with others from various interest groups, their implementations in this regard differ. For instance, 365Key displays a list of interest groups for users to choose from while Del.icio.us requires its users to create their own networks or locate interest groups of relevance by tags or registered user names. Furthermore, Del.icio.us integrates tagging with other Web 2.0 applications (e.g., blogs and Facebook). In contrast, 365Key allows its users to add its logo to other websites of their choices.

Content Analysis of Tags

As described in the Data Collection and Analysis section, 818 and 471 unique tags were gathered from Del.icio.us and 365Key respectively. Table 2 lists some parameters of the tags extracted from each site. It seems that Chinese taggers tend to use identical words to tag news items, which can be partially explained by the nature of its tagging organization system. Some 365Key taggers also use tags 365Key supplies rather than choose on their own. These two practices of 365Key taggers perhaps account for why they created fewer unique tags than their counterparts at Del.icio.us. It must be pointed out as well that on average Del.icio.us users created five tags per tagged item while 365Key taggers only did three.

Table 2. Some Parameters of Tags
 Tags CreatedRecords TaggedUnique TagsPercentage
Del.icio.us5064100081816.2
365Key1300644104713.62

Despite the different tagging statistics presented in Table 2, tag frequency distributions of both sites look quite similar (See Figure 1). Taggers seem often using broad, general terms (e.g., blog) rather than specific, less general words (e.g., blogspot) when tagging a same item. The well-known 80–20 rule appears applicable in tagging too according to the results of this study.

thumbnail image

Figure 1. Tag Distribution at Del.icio.us and 365Key (X-axis: Tag Ranking, Y-axis: Tag Frequency)

Download figure to PowerPoint

As mentioned earlier, tags with a usage frequency of 10 or above at both sites were chosen for content analysis. The rationale for selecting the number 10 as the threshold is based more on feasibility than on any other particular reasoning. This threshold reduces the number of unique tags at Del.icio.us to 89 and 59 for 365Key. The following content analysis was thus performed on those two sets of tags from Del.icio.us and 365Key respectively.

Tag Types

The afore-described two sets of tags were first categorized according to their parts of speech (e.g., noun, adjective). Table 3 presents the results of such analysis.

Table 3. Tag Types–Parts of Speech
Thumbnail image of

As shown in Table 3, noun tags are used most often at both sites, which should be the case because tagging essentially resembles to keyword indexing and appropriate candidate terms for tags or keywords should be nouns or noun phrases. However, taggers often choose adjectives to express their feelings and emotion (e.g., Kipp 2007a) while indexers are trained not to do so in indexing. That is why several adjectives or even verbs were chosen by taggers as shown in Table 3. This finding in fact echoes what Kipp (2007a) found by dividing the tags she collected into two groups: affective tags and time/task related tags. She indicated that most affective tags turned out to be adjectives. In contrast, nouns would commonly be candidates for time/task related tags. Howerver, there are still several words in our study that would not fall into neither group Kipp created, and one such instance is “military”.

A further analysis of the tags collected in this study reveals more characteristics with regard to tag types (See Table 4). 365Key taggers often use English for IT terms. IT actually is ranked as one of the most popular tags in China. By comparison, Del.icio.us taggers rarely use terms from other languages. Some linguistic problems become more obvious in the tagging process at Del.icio.us than at 365Key because they all are intrinsic to the English language. Polysemy (e.g. Apple, feeds), synonyms (e.g. movies vs. film) and plurals (e.g. game vs. games) are some examples of such problems (Noruzi 2006). At Del.icio.us, for instance, the word “blogs” was used 177 times for tagging while “blog” was also chosen 49 times out of all the tags gathered in this study. Furthermore, it appears interesting to note that net-born terms, e.g., howto, feeds, aggregator at Del.icio.us; (joking), MM (beauty), (tabloid news), (unintended revealing) at 365Key, often used in text messaging, chat rooms and the like, are ranked high at both sites in terms of frequency.

Table 4. Tag Types – Other Characteristics
 Foreign TermsTerm VariantsNet-born Terms
Del.icio.us023
365Key604

Beyond Tag Types

In the previous section, we analyzed the two sets of tags by their types. We intend to explore below the specific terms taggers used at both sites. Table 5 lists terms of same meaning by tagging rank in the two groups of tags chosen for this analysis. According to Table 5, it becomes obvious that “news” is the only tag that achieved the same rank yet the leading place out of all the tags extracted from both Del.icio.us and 365Key. The reason why we obtained this prominent result is because all the tags analyzed in this study were extracted from the News category at the two chosen tagging sites.

Table 5. Tags of Same Meaning
Thumbnail image of

In addition, some terms actually represent identical concepts although their spellings may turn out otherwise. For example, “IT” at 365Key, as a tag, is close in meaning to “technology” or “Web” at Del.icio.us because “Web” is one kind of “IT” while “technology” is a broader term of “IT”. Tags of same meaning are usually terms with broad, general connotation. It seems that 365Key taggers normally choose terms relating to social life such as (entertainment) and (culture) while users at Del.icio.us often tag from the perspective of science and technology.

In a more focused manner, the 12 top ranked tags from both sets were further analyzed and Table 6 presents the results. As illustrated in Table 6, 365Key users appear in favor of general tags, e.g., (life) while Del.icio.us taggers seem interested in using specific terms (e.g., daily). Moreover, Del.icio.us users tend to tag the source (e.g., blog, Web) and other dimension (e.g., design) of news whereas 365Key taggers often group news into one category (e.g., finance) with general tags. This finding, in a sense, reflects the different cultures of China and the USA. Chinese customarily describe things in an abstract and implicit manner while Americans, taking a straightforward approach, often express their views and thoughts explicitly.

Table 6. An Analysis of 12 Top Ranked Tags
Thumbnail image of

Conclusions

  1. Top of page
  2. Abstract
  3. Introduction
  4. Related Research
  5. Data Collection and Analysis
  6. Results and Discussion
  7. Conclusions
  8. Acknowledgements
  9. References

Indeed, tagging in China and the USA on one hand shows some differences and, on the other, shares many similarities. Although no surprises were found in this study with regard to tagging in two different countries, several points can be made based on what has been presented in this report.

First of all, 365Key—the Chinese tagging site - provides its users with more pre-set functions whereas Del.icio.us gives more freedom to its taggers. Second, tagging, like many other information behaviors, is greatly influenced by and stamped with the social and cultural traditions existing in each country. Third, the tags of same or similar meanings chosen by both groups of taggers indicate that tagging, regardless of where it is done and where the tagger is from, is basically performed in accordance with the fundamental rules in indexing (e.g., nouns and noun phrases as tags). Four, tagging is NOT keyword indexing—one of the major methods for information representation—that has been done in producing database systems (e.g., InfoTrac) and search engines (e.g., Google). Rather, tagging is an activity unique in the networked environment for loosely representing and organizing all kinds of information. Whether tagging will be seriously and successfully adopted and incorporated into other environments for information retrieval purpose (e.g., OPACs) is a topic beyond the scope of this study.

Needless to say, the current research only touches upon the tip of the gigantic “tagging” iceberg by analyzing two tagging sites in China and the USA. Further study on this intriguing theme is planned. We are looking forward to this opportunity as well as challenge.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Related Research
  5. Data Collection and Analysis
  6. Results and Discussion
  7. Conclusions
  8. Acknowledgements
  9. References

One of the authors, Chen Xu, gratefully acknowledges the funding from the China Scholarship Council, which makes it possible for her to conduct her doctoral research in the USA for two years. The support of Professor Feicheng Ma and his research team at School of Information Management, Wuhan University, China during the course of this study is greatly appreciated. We also thank Professor Margaret E.I. Kipp at Long Island University for her kind support and helpful suggestions.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Related Research
  5. Data Collection and Analysis
  6. Results and Discussion
  7. Conclusions
  8. Acknowledgements
  9. References
  • Binkowski, P.I. (2006). The effect of social proof on tag selection in social bookmarking applications. Master's thesis. Chapel Hill, NC: School of Information and Library Science, University of North Caroline at Chapel Hill. http://etd.ils.unc.edu/dspace/handle/1901/358.
  • Farooq, U., Kannampallil, T.G., Song, Y., Ganoe, C.H., Carroll, J.M., & Giles, C.L. (2007). Evaluating tagging behavior in social bookmarking systems: Metrics and design heuristics. Proceedings of the ACM Conference on Supporting Group Work, (pp. 351360). Sanibel Island, Florida. Novbember 4–7, 2007.
  • Fitchter, D. (2006). Intranet application for tagging and folksonomy. Online, 30 (3), 4345.
  • Golder, S. A., and Huberman, B. A. (2006). Usage patterns of collaborative tagging systems. Journal of Information Science, 32 (2), 198208.
  • Hong, B. (2005). Viewing frequency and bookmarking frequency. Blog post (July 30, 2005): http://blog.donews.com/keso/archive/2005/07/30/487503.aspx.
  • Kipp, M.E.I. (2007a). @toread and cool: Tagging for time, task and emotion. Proceedings of the 8th Information Architecture Summit. Las Vegas, Nevada, March 22–27, 2007. Also available: http://eprints.rclis.org/archive/00010445/01/mkipp-iasummit2007.pdf.
  • Kipp, M.E.I. (2007b). Tagging for health information organisation and retrieval. Proceedings of the Joint Conference on Digital Libraries, Vancouver, British Columbia, June 18–23, 2007. Also available: http://eprints.rclis.org/archive/00011411/.
  • Kipp, M.E.I. (2006) Complementary or discrete contexts in online indexing: A comparison of user, creator, and intermediary keywords. Canadian Journal of Information and Library Science. (In press). Also available: http://eprints.rclis.org/archive/00008379/.
  • Kipp, M.E.I., and Campbell, D.G. (2006). Patterns and inconsistencies in collaborative tagging systems: An examination of tagging practices. Proceedings of the Annual Meeting of the American Society for Information Science and Technology, Austin, Texas, October 19–24. Also available: http://eprints.rclis.org/archive/00008315/.
  • Muller, M.J. (2007). Comparing tagging vocabularies among four enterprise tag-based services. Proceedings of the ACM Conference on Supporting Group Work, (pp. 341350). Sanibel Island, Florida. Novbember 4–7, 2007
  • Noruzi, A. (2006). Folksnomies: (Un)Controlled vocabulary. Knowledge Organization, 33 (4), 199203.
  • Rainie, L. (2007). Interview: Author David Weinberger describes how tagging changes people's relationship to information and each other. Pew Internet & American Life Project. http://www.pewinternet.org/pdfs/PIP_Tagging.pdf.
  • Vander Wal, T. (2004). You down with folksonomy? Blog post (August 5, 2004): http://www.vanderwal.net/random/entrysel.php?blog=1529.
  • Vander Wal, T. (2005). Explaining and showing broad and narrow folksonomies. Blog post (February 21, 2005): http://www.vanderwal.net/random/entrysel.php?blog=1635.