Tweet, tweet … analysing a library conference backchannel with Hawksey’s TAGS

Twitter has gained a reputation as a social media tool which is very popular within the LIS community, and most libraries and archives, LIS schools, and library/information conferences, and well as many individuals in the discipline and profession, make serious use of it for information exchange. Being able to easily get an analysis of the tweets around a topic is therefore very useful for LIS folk, as well as giving an insight into the increasingly important area of social media data analysis.

In this post, I give an account of how one tool, Hawksey’s TAGS, can be used in this way. As someone who had never used this, or any similar system, before, I was particularly impressed that I was able to get useful analysis within 20 minutes from starting cold.

Devised as a “hobby” by Martin Hawksey, TAGS is a free Google Sheet template which gives simple automated collection and display of Twitter search results. It uses the Twitter API to identify tweets according to specified criteria, creates an archive of tweets, and uses Google’s visualisation tools to display them. It is very easy to use, but if need be support and help is provided through online forums.

To use TAGS in a simple way – and there are more complicated and clever things that can be done with it, but that’s for another day – all that is necessary is to (1) get a Google account, if you do not already have one, (2) download TAGS (6.1 is the current version) into your Google Drive, and (3) get a Twitter authorisation to collect Twitter data (the TAGS software prompts you to do this, and it should only be necessary to do it once). Then for each analysis you wish to do, you run TAGS specifying the collection criteria (essentially entering terms into what is effectively a search box), and wait for the script to complete. Basic statistics on the number of tweets included, and the time period overview which they were sent, are presented. You can then make the the archive usable, by clicking on the ‘share’ button, and visualise it by selecting TAGSExplorer. (You can also make the archive searchable for detailed analysis, but that is also for another day.)

We can exemplify this by looking at tweets about CILIP’s annual conference held in July 2017, which used the hashtag ‘CILIPConf17’. The screenshot below shows TAGS set up to this search; it is started by clicking ‘TAGS’ and ‘Run Now’.

The simplest way of using TAGS is to find all Tweets with a particular characteristic, for example including the CILIP conference hashtag. A partial display of these is shown below, from an analysis done by my CityLIS colleague Lyn Robinson. The somewhat indigestible display below does give a good impression of the extent of twitter activity. The linking lines show Twitter users replying to one another; the isolated user names are those who use the hashtags, but do not engage. The size of the usernames indicates the frequency of replies; there is, however, no distinction between the isolated users based on the number of tweets. This display is gives an indication of who is using the hashtags, but not to what extent, and how much conversation is taking place. At the left-hand side is shown a list of the most common hashtags in this tweet archive: not surprising ‘#CILIPConf17’ (which had to be in all tweets) is at the top, followed by two more CILIP-related tags, ‘#factsmatter’ and ‘#CILIPethics’; next in line is the ‘#CityLIS’ tag, showing the engagement of CityLIS tweeters with the conference.

The display below shows tweets about the CILIP conference which also mention the CityLIS library school; it is created simply by entering ‘#CILIPConf17’ and ‘#citylis’ as the search term.

The same can be done using a twitter username, as in the display below, created by entering ‘@floridi’ and ‘#CILIPconf17’ to find tweets mentioning Luciano Floridi who gave a keynote talk. Again the lines join those who interact, the size of the name showing the extent of interaction, with the isolated usernames being those who simply mention Floridi.

More extensive Boolean logic can be used. For example, the display below shows tweets using the ‘CILIPConf17’ with either ‘#citylis’, ‘@ucldis’ or ”@infoschoolsheff’, to find tweets from, or mentioning, any of three leading UK LIS departments. Note that there are limitations to the complexity of Boolean logic can be employed (because of the way Google interrogates Twitter, rather than limitations in the TAGS software) but simple combination of ANDs and ORs work fine.

It is important to remember that TAGS is intended as a simple, quick and free tool, and therefore has some limitations. Some issues in the way the visualisation works have been noted. There are also limitations in the collection of tweets, using the Twitter API. As Twitter themselves say “it’s important to know that the Search API is focused on relevance and nor completeness. This means some Tweets and users may be missing from search results.”

Also, TAGS only accesses tweets from roughly the previous week (between 6 and 9 days), because of the limitations on the length of time that tweets remain available; there are ways round this, but they are more complicated that the simple use of TAGS. There are also limitations on the numbers of tweets that TAGS will handle, though this will only affect analyses of topics with very large volumes.

So TAGS are best used to get a quick and simple picture of recent Twitter activity in a topic of interest. If completeness and precision are required, then it will be necessary to extra work in identifying a full set of Tweets, and then checking and cleaning the data. This might involve collecting tweets over a long period, or exporting the data to a more sophisticated analysis and visualisation program such as NodeXL. For an example of the use of NodeXL, for detailed analysis see the paper by Lee et al. on an analysis of a dataset of Tweets from three annual conferences of the Association of Internet Researchers, showing the nature of the networks formed, the most influential Tweeters, and the topics mentioned. This paper also gives a good literature review of examples of Twitter analysis.

Bear in mind that Twitter has rules about collecting Twitter data sets and making them public, and that these rules change to match new uses; see the current Twitter terms; however, it is unlikely that they would be infringed by the kind of small scale analysis and display exemplified by this CILIP conference example.

The TAGS software allows simple but effective Twitter analysis to be done with very little effort or resources, and is often all that is needed. It is something that anyone from an LIS background interested in how Twitter is being used should get familiar with.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.