Understanding Twitter use by major LIS professional organisations in the United States

Abstract

Although Twitter has been widely adopted by professional organisations, there has been a lack of understanding and research on its utilisation. This article presents a study that looks into how five major library and information science (LIS) professional organisations in the United States use Twitter, including the American Library Association (ALA), Special Libraries Association (SLA), Association for Library and Information Science Education (ALISE), Association for Information Science and Technology (ASIS&T) and the iSchools. Specifically explored are the characteristics of Twitter usage, such as prevalent topics or contents, type of users involved, as well as the user influence based on number of mentions and retweets. The article also presents the network interactions among the LIS associations on Twitter. A systematic Twitter analysis framework of descriptive analytics, content analytics, user analysis and network analytics with relevant metrics used in this study can be applied to other studies of Twitter use.

Keywords

ALA ALISE ASIS&T iSchools LIS professional organisations SLA Social media outreach Twitter analysis Twitter use

1. Introduction

Social networking sites, also known as social media, are platforms allowing users to create personal profiles and to connect and interact with other users in real time. Examples of such social networking sites include Facebook, Instagram, Flickr and Twitter. Launched in 2006, Twitter has become one of the fastest growing and widely used of these services, free of charge to any individual or group with their account tied to an email address. Twitter allows users to communicate using short messages, up to 140 characters, called ‘tweets’. These messages can contain text, URLs and media such as photos, audio and video. Users can repost others’ messages as ‘retweets’ (RT) or modified ‘retweets’ (MT). Public messages are searchable on the service and can be thematically connected by topical terms called hashtags (#) and directed to users with username mentions (@). This allows users to connect topics and construct conversations with others who share their interests, without necessarily having personal relationships. Given these features, Twitter has become an effective and cost-efficient marketing tool.

Like other fields, library and information science (LIS) professionals are creating institutional and individual Twitter accounts for marketing, providing and promoting patron services and other professional communications. Affordable and easy to manage, Twitter provides new opportunities for libraries and institutions to reach out to their patrons directly and instantly; it easily conveys information about the library’s hours, events, services and resources [1,2]. Twitter has been used for social networking between libraries and patrons and among information professionals [3,4]. Professional organisations in the field of LIS use Twitter in a similar fashion. However, such use is still a relatively new area for study, unlike use by individuals, businesses and libraries. This study explores the use of Twitter by major LIS professional organisations in the United States. The findings provide insights on how LIS organisations use Twitter and can help librarians and information scientists better understand the trends and the use of Twitter between major LIS organisations. As one of few studies first analysing Twitter use by professional organisations with multiple analysis approaches (descriptive analysis, content analysis and network analysis), the research design used in this article provides perspectives for the LIS research community in a broader sense.

2. Literature review

2.1. The role of professional associations

A professional association is a body of persons engaged in the same profession, formed usually to control entry into the profession, maintain standards and represent the profession in discussion with other bodies [5]. The LIS profession is supported by a diverse range of professional associations, with each of these organisations providing services and resources for information professionals and advocacy and leadership for the development and delivery of library and information services [6]. LIS associations exist at the international, national, regional, state and local levels; these groups all support practitioners and members with needs, ranging from general to specific, in the area of LIS [7]. This article focuses on the national-level organisations in the United States.

In general, the goals of these associations include the maintenance and development of members’ interests and professional skills met through offering continuing professional development, training research and education. Other objectives include the dissemination of professional news and up-to-date research and trends, the provision of networking and communication opportunities among members and the organisation and providing advocacy for the profession. LIS organisations face many challenges while attempting to reach these goals. According to Madden’s review [8], six common challenges that LIS organisations face include the following:

Attracting and retaining members [9];

Marketing and promoting services [10];

Obtaining and generating funding [11];

Keeping information and research up-to-date for members [12];

Assessing the effectiveness of provided services [13];

Developing information policy and strategies [14].

Twitter is a powerful tool for disseminating information and building online communities for any organisation and is expected to be an effective platform for LIS associations to accomplish such primary goals as providing members with networking and professional development opportunities and in advocating for the profession.

2.2. Twitter use in LIS

Given the increasing popularity and adoption of social networking sites by libraries and institutions, social media studies have become a field of interest for researchers and LIS practitioners. According to Singh and Gill [15], there were more than 200 articles on Web 2.0 technology (including social networking sites and blogs) published in 13 leading LIS journals between 2007 and 2011, making it a trending topic for a growing number of publications. These studies explored the implementation of social media tools, including Twitter, by libraries in various contexts (i.e. public libraries, academic libraries and academic institutions) and from various angles, including why and how libraries use social media [16 –20], the impact of social media [21], identifying connecting users [22,23] and social media policies and guidelines [24].

Twitter is ‘social’ media, not a one-way marketing tool. It can be used to learn about customers/patrons and for getting feedback through their tweeted ‘conversations’. As indicated earlier, a growing coterie of professional organisations and libraries use Twitter in a variety of ways.

2.3. Research tools for Twitter studies

As Twitter studies become an emerging research area, data collection and analysis methods are evolving. In general, there are three major sources for obtaining Twitter data: Twitter search, Twitter application programming interfaces (API) and commercial vendors:

Twitter search. Researchers can conduct searches on the site (http://twitter.com/search) using keywords or hashtags. It allows you to retrieve up to 7 days of historical data or 1500 tweets [25]. These search results can be copied manually or downloaded using an online add-on application such as NCapture, a web browser extension to capture content like web pages and social media for analysis in NVivo.

Twitter API. Researchers can use programmes to automatically retrieve Twitter data directly from the site through an API. There is also a plug-in with NodeXL for Microsoft Excel that allows users to explore network graphs. However, according to Kim et al. [25], the number of tweets retrievable through this method is capped at approximately 1% of all tweets.

Commercial vendor. Researchers can also purchase full Twitter data, including historical data, through commercial vendors who also provide applications that allow for real-time analysis of data. Each source presents unique challenges and values (for comparison of cost and attributes, see Kim et al. [25]).

Previous studies have used a number of analytical methods to gain a better understanding of Twitter data. However, common Twitter analytical frameworks include descriptive analysis, content analysis and network analysis [26]. While descriptive analysis is essential to give an overview of a data set, content analysis is the most common method used in Twitter studies [27]. Given the text-based, user-provided nature of Twitter data, content analysis naturally becomes one of the major analytic methods, allowing researchers to investigate research questions such as user involvement, topics discussed and types and intentions of use. It allows researchers to examine trends, characteristics, patterns, the typology of content in various perspectives and so on. Content analysis can be carried out manually (manual coding based on a set of classifications/themes) or automatically (data mining) by software. Similarly, as Twitter is one of the major social networking sites, network analysis logically suits research needs and becomes another major analytical method to examine the often large data sets generated by users directly, such as their contributed tweets, as well as indirectly, such as the networks formed by their interactions on Twitter. It allows researchers to visualise and reveal the interactions and dynamics between users.

2.4. Research questions

As discussed, the Twitter analytics framework combines three primary analytic techniques, each intended to examine data from different metrics. This study follows this analytics framework to explore the Twitter use of five national-level LIS professional associations based in the United States. This study specifically explores the characteristics of these organisations’ Twitter usage and interactions and is driven to answer the following research questions:

What are the characteristics of LIS association tweets?

What topics or concerns do these LIS association tweets share?

Who are the Twitter users involved in the LIS associations’ tweets?

What are the interactions among the LIS associations on Twitter?

3. Method

3.1. Sample of LIS professional organisations

Five major US LIS professional organisations included in this study are as follows:

American Library Association (ALA);

Special Libraries Association (SLA);

Association for Information Science and Technology (ASIS&T);

Association for Library and Information Science Education (ALISE);

The iSchools, a consortium of information schools.

The first four are well-established organisations that represent a broad range of the LIS field [28]. ALA is the largest and oldest library association in the world; SLA serves information professionals working in such settings as corporations, the law, academic institutions and government; ASIS&T serves both information science researchers and professionals alike; and ALISE focuses on LIS education. Although established in 2005, iSchools has emerged as a major force for advancing the information field in the 21st century. All five organisations are active on Twitter. It should be noted that although these organisations were initially incorporated in the United States, over time, they have extended their memberships worldwide.

3.2. Data collection and analysis

The researchers utilised NCapture to capture data from Twitter accounts. NCapture (a web browser extension) was used to obtain tweets from the Twitter accounts of these five LIS organisations every 1 or 2 days during a 2-month period from 12 October to 12 December 2015, capturing a total of 15,518 tweets. Table 1 shows a range of basic descriptive statistics of the sample data set, including the number of tweets, distribution of different types, number of hashtags and number of tweets containing URLs and so on.

Table 1.

Number of tweets captured from five LIS associations and basic descriptive statistics

Metric	ALA	SLA	ASIS&T	ALISE	iSchools	Total
Number of followers	75,400	4488	4668	586	2927	88,069
Number of tweets in account history	14,000	11,500	7286	1465	3781	38,032
Number of all types of tweets captured in the sample (see breakdown in next two rows)	3330	4041	3360	1462	3325	15,518
Number of original tweets	2024	2744	2844	647	1053	9312
Number of retweets	1306	1297	516	815	2272	6206
Number of tweets with hashtags	2164	2122	1712	790	1802	8590
Number of hashtags	790	279	619	368	735	2354 (unique)
Number of tweets with links	2617	3448	2692	965	3045	12,767
Number of tweets with mentions	1369	1517	2058	598	1336	6878
Account creation dates	February 2009	March 2009	August 2009	December 2012	January 2009
Time frame coverage of tweets in sample	14 April 2014–12 December 2015	10 April 2015–12 December 2015	6 June 2013–11 December 2015	5 December 2012–11 December 2015	1 December 2010–10 December 2015

LIS: library and information science; ALA: American Library Association; SLA: Special Libraries Association; ASIS&T: Association for Information Science and Technology; ALISE: Association for Library and Information Science Education.

Twitter data are primarily texts, ‘unstructured’ or ‘semi-structured’ in nature and composed of a short list of words, hashtags, URLs and other information. It is necessary to use content analytics, which refers to a broad set of natural language processing and text mining methods, for extracting intelligence from Twitter data. This study conducted a content analysis of themes and hashtags to explore the subject matter of LIS professional organisations’ tweets.

NVivo 11 Pro (NVivo for short, subsequently), the latest version of a qualitative and mixed-methods data analysis tool, was used for processing and analysing data. Data from the preliminary descriptive statistical analysis using NVivo were also converted to a Microsoft Excel spreadsheet and imported to Access for further analysis. For qualitative analysis, summative content analysis began with identifying themes and hashtags. Major topics were identified by a combination of automatic text mining using NVivo’s auto-themes analysis function and manual categorisation by two authors of this article who followed a two-step approach. They first conducted independent categorisation of the NVivo-generated themes and then met to resolve any differences until all categories were agreed upon as consistent.

The usernames specifically mentioned and retweeted by the five organisations were extracted from tweets to identify the ‘most visible’ or ‘influential’ users. Data were exported in comma-separated values (CSV) format from the spreadsheet and then imported into the Gephi programme in order to visualise the social network embodied in the data.

3.3. Limitations

The intended tweets to be examined were those that originated from or were retweeted by the associations’ accounts. Ideally, all tweets associated with each account should be included. However, the number of tweets that can be captured and the time frame in a given sample are randomly determined by Twitter, and the function of the Twitter application itself, for the account-based data capture request. Variables may include the amount of user traffic and the number of available tweets for an account when captured. As shown in Table 1, the number of ALA tweets in the sample accounts for 24% of that account’s total tweets, while the tweets captured for ALISE account for nearly 100% of their total. Despite this limitation, given the exploratory and descriptive nature of this study, we believe that using the maximum number of captured tweets from all accounts would be more helpful to shed light on the Twitter use of these organisations as a whole, rather than a smaller subset of tweets.

4. Results

4.1. Descriptive analytics

As shown in Table 1, among the 15,518 tweets, 60% (9312) were original and 40% (6206) were RT. Over 55% (8590) of tweets contained at least one hashtag, and 2354 different hashtags were captured. The ratio of an account’s tweets containing at least one hashtag ranged from 54% (iSchools) to 65% (ALA). This result suggests that most of these tweets were topical in nature. URLs were popular tweet content, with 82% of sampled tweets containing at least one. Here, the ratios ranged from the highest of 92% of iSchools’ tweets to the lowest 66% of ALISE’s. In addition, 44% of all tweets collected mentioned at least one user. ASIS&T had the most conversational or interactive tweets, with 61% of theirs containing usernames. Conversely, SLA appeared the least conversational or interactive, as only 38% of their tweets mentioning users.

4.2. Themes analysis by tweet contents

4.2.1. The theme categories of five LIS professional organisations’ tweets

The auto-themes analysis function in NVivo extracted 20 themes from the imported sample data. These themes were grouped into seven categories, as judged by the keywords and phrases included under each, by two authors, in order to avoid subjectivism and prejudice. Table 2 summarises the category distribution:

The largest category, Libraries (all types) & Services, includes themes such as public and school libraries, library users and usage, book recommendations and hours of operation.

Research is the second largest category, mainly discussing papers/white papers, research (LIS, academic etc.) and data (big data, data science etc.).

Conferences/Webinars/Continuing Education themes include conferences, posters, registration, sessions and webinars.

Information concepts is a diverse category, including information literacy and visibility.

LIS Education consists of references relating to students (those attending iSchools, student groups etc.) and online courses.

Librarians and Jobs includes career-related professional references.

Social Media is the smallest category and includes Web 2.0 resource-related themes such as links to blogs and other social media sites.

Table 2.

The theme categories of five LIS professional organisations’ tweets

Theme category	Frequency	Percentage
Libraries (all types) & Services	877	27
Research	546	17
Conferences/Webinars/Continuing Education	528	16
Information Concepts	374	12
LIS Education	358	11
Librarians and Jobs (all LIS career/job related)	345	11
Social Media (Web 2.0 resource related)	196	6

LIS: library and information science.

Using the matrix coding by NVivo, data on the number of themes included in these organisations’ tweets were extracted, imported into an Excel spreadsheet and then plotted. Certain distribution characteristics were observed: the primary ALA theme was Libraries (all types) & Services, SLA focused more on ‘Librarians and Jobs’ and ‘Conferences/Webinars/Continuing Education’, ASIS&T’s predominant themes were ‘Information concepts’ and ‘Research’ and iSchools focused on ‘LIS Education’. ALISE did not show any obvious thematic preferences.

4.2.2. Tweet content cluster analysis by account

In order to explore the content similarity between these five LIS professional organisations’ tweets, a cluster analysis was conducted using NVivo. Figure 1 shows the results. ALISE and iSchools are clustered into one group largely due to their educational nature with more LIS education–related themes. SLA, ALA and ASIS&T are clustered into another group because of their professional orientation, with the latter two showing more common interests in LIS practice and research. SLA pays unique attention to specialty and personalised topics such as career development.

Figure 1.

Source clustered by content similarity

4.3. Hashtag analysis

Hashtags are a means of organising and making trending discussion topics discoverable on Twitter. Contemporaneous events were identifiable through hashtag analysis.

4.3.1. Characteristics of hashtag use

A total of 55% (8590) of the 15,518 tweets included at least one of the most frequently used 2354 unique hashtags found in analysis. However, only four hashtags, #libraries, #library, #copyright and #Twitter, were used by all five organisations. These most commonly used hashtags accounted for 0.17% of total use. In all, 31 (1.3%) hashtags were used by four organisations. In total, 2062 hashtags were unique to only one user organisation, accounting for 87.55% of all hashtags. This demonstrates that only a few, extremely popular topics were commonly mentioned, and most hashtags were unique to only one organisation. SLA had the lowest ratio of hashtag uniqueness at 13%, while ALISE scored the highest, with 47% of their hashtags being unique.

4.3.2. Most commonly used hashtags by the five LIS professional organisations

The 35 most commonly used hashtags, utilised by at least four organisations, were extracted in order to observe which popular topics and events were being discussed. These hashtags were classified into eight categories: libraries and services, conferences, LIS education, data-related, librarians, information policy and information literacy, social media and technology. Further investigation revealed that the hashtags #libraries, #library, #archives, #librarians, #librarian, #bigdata, #socialmedia and #lis were each mentioned at least three times, representing topics of focus for all five organisations. Along with general topics regarding the development of LIS practice and theories, popular specific topics like big data, social media, massive open online courses (MOOC) and three-dimensional (3D) Printing were widely discussed.

Each organisation exhibited a particular distribution of hashtags, as shown in Table 3. ALA used the greatest number of them, and SLA the fewest. From the subject categories, it could be observed that each LIS organisation had specific thematic focal points: ALA focused on topics about library and user services, librarians, privacy and netneutrality; iSchools and ASIS&T mostly showed concern for data and social media. ALISE, ASIS&T and iSchools paid more attention to LIS education. ALISE also showed interest in #FollowFriday, a popular account recommendation model on Twitter. #MOOC made for a heated SLA discussion.

Table 3.

Distribution of hashtags used by at least four LIS professional organisations

Category	Hashtags	ALA	SLA	ASIS&T	ALISE	iSchools	Total
Libraries & Services	libraries	189	9	34	25	48	305
	library	77	18	15	14	18	142
	archives	5	0	15	9	6	35
	Books	7	0	1	1	5	14
	Imlsfocus	2	0	1	3	1	7
	IMLS	1	0	1	1	1	4
	FF	6	0	6	31	2	45
	tlchat	4	0	3	3	1	11
Conferences	ALISE2015	1	0	57	189	1	248
	ALAac14	138	0	2	5	2	147
	ALAAC15	114	2	0	4	1	121
	NLW15	29	2	0	2	1	34
LIS Education	lis	0	4	42	58	23	127
	MOOC	1	25	5	0	9	40
	highered	1	0	9	10	13	33
	gslisui	1	1	0	3	27	32
	education	2	0	7	1	1	11
Data-related	bigdata	0	4	29	3	60	96
	datascience	1	0	18	2	17	38
	dataviz	0	2	14	2	8	26
	openData	1	2	14	0	8	25
	Data	1	0	3	4	8	16
Librarians	librarians	50	12	3	0	14	79
Librarians	librarian	4	5	3	0	4	16
Information policy and information literacy	copyright	1	6	2	2	3	14
	netneutrality	27	0	3	6	2	38
	privacy	9	0	3	1	3	16
	infolit	0	1	1	2	2	6
Social media	Twitter	2	1	1	2	3	9
Social media	socialmedia	3	0	25	3	13	44
Technology	mobile	0	3	4	1	3	11
	makerspaces	4	0	1	1	2	8
	digital	1	0	2	1	1	5
	Curation	1	0	1	2	1	5
	3dprinting	2	1	0	1	1	5
Total		685	98	325	392	313	1813

4.3.3. Top 30 most frequently used hashtags by five LIS professional organisations

After sorting hashtags by the number of occurrences in tweets, two individual coders extracted and classified the 30 most frequently used hashtags into six categories: conferences, libraries and services, LIS education, librarians and jobs, big data and communication. It was noted that the hashtags #libraries, #library, #librarians, #bigdata and #lis all received three mentions by at least four organisations. In all, 10 out of the top 30 hashtags regarded conferences held by the five organisations, and the rest were general topics of LIS practice, development and education.

Each organisation had preferred topics, as shown in Table 4. ALA focused on #library, #libraries, #librariestransform, #libraryofthefuture and promoted reading. SLA used #lisjobs and #BeRevolutionary while emphasising interaction with users. iSchools concentrated on member information. ASIS&T facilitated discussions on various topics. For example, #10MinReads introduced many new and interesting ideas, including the history of American maps and data literacy. ALISE highlighted their conferences.

Table 4.

Top 30 most frequently used hashtags

Category	Hashtags	ALA	SLA	ASIS&T	ALISE	iSchools	Total	Number of LIS associations
Conferences	SLA2015	0	1118	0	1	0	1119	2
	ALISE2015	1	0	57	189	1	248	4
	alamw15	153	0	2	1	0	156	3
	ALAac14	138	0	2	5	2	147	4
	ALISE2014	0	0	0	138	2	140	2
	ALAAC15	114	2	0	4	1	121	4
	ias15	0	0	120	0	0	120	1
	asist2013	0	0	115	1	0	116	2
	asist2014	0	0	97	2	0	99	2
	asist2015	0	0	81	1	1	83	3
Libraries & Services	libraries	189	9	34	25	48	305	5
	library	77	18	15	14	18	142	5
	librariestransform	120	0	0	0	0	120	1
	libraryofthefuture	108	0	0	0	1	109	2
	BeRevolutionary	0	104	0	0	0	104	1
	ALA_Carnegie	119	0	0	0	0	119	1
	ALAYMA	65	0	0	0	0	65	1
LIS education	Heinz	0	0	0	0	265	265	1
	Syracuse	0	0	0	1	132	133	2
	lis	0	4	42	58	23	127	4
	InfoSpace	0	0	0	1	124	125	2
	UWash	0	0	0	0	95	95	1
Librarians & Jobs	librarians	50	12	3	0	14	79	4
	lisjobs	0	187	85	1	0	273	3
	LibJobs	0	188	6	1	0	195	3
	SLAIETLIP	0	86	0	0	0	86	1
Communication	SLAtalk	0	434	0	0	1	435	2
	10MinReads	0	0	157	1	0	158	2
	TeamALA	68	0	0	0	1	69	2
Big data	bigdata	0	4	29	3	60	96	4
Total		1202	2166	845	447	789	5449

4.4. Twitter user analysis

‘Influential’ and highly visible accounts are identified through others mentioning and retweeting them [29,30]. Retweeted content is typically identified using ‘RT (or MT) @username’ within the text, while user mentions were identified by searching tweets for @usernames (without ‘RT’ or ‘MT’). This study focused on users retweeted and mentioned by the five LIS professional organisations. Table 5 shows the number of unique tweet users referenced by the five LIS associations. A total of 1866 unique users were retweeted and 3329 were mentioned. In the sample data, two types of users were identified: individual users and institutional users. They were analysed separately in section 4.4.1.

Table 5.

Number of tweet users involved in the LIS associations’ tweets

Metric	ALA	SLA	ASIS&T	ALISE	iSchools	Combined
Number of retweeted users	632	327	317	409	433	1866 (unique)
Number of mentioned users	842	460	956	572	973	3329 (unique)

4.4.1. Users retweeted by five LIS professional organisations

Out of the 1866 unique users, only 3 (0.16%) of them were retweeted by all five organisations. In all, 9 (0.48%) were retweeted by four organisations, and 38 (2%) were retweeted by three organisations. Two organisations retweeted 137 (7.3%) of these users. In all, 1679 (89.99%) of these users were retweeted by just one organisation. This demonstrates that most users are exclusively retweeted by each organisation, while very few capture the interest of all five organisations.

The top 10 most commonly and most frequently retweeted users by the five LIS associations are shown in Table 6. The top 10 most commonly retweeted institutional users list mainly consists of other LIS professional organisations (@CILIPinfo, @ASLIBRARIES, @IFLA), branches of LIS associations (@amlibraries, @yalsa, @alscblog, @ALAnews) and iSchools members (@umsi, @ISchooIsU, @InfoSchoolSheff). These accounts were each retweeted an average of 72 times by up to four of the organisations.

Table 6.

Top 10 most retweeted institutional users

Metric	Username	Frequency	Number of LIS organisations
Most commonly retweeted	CILIPinfo	43	5
	amlibraries	35	5
	iSchoolsU	296	4
	yalsa	36	4
	alscblog	32	4
	ALAnews	9	4
	ASLIBRARIES	6	4
	IFLA	6	4
	umsi	137	3
	InfoSchoolSheff	119	3
Most frequently retweeted	iSchoolsU	296	4
	UW_iSchool	151	2
	umsi	137	3
	SISCSU	125	2
	InfoSchoolSheff	119	3
	SLANewEngland	106	2
	iusoic	105	2
	iconf	80	2
	gslis	79	2
	SLAconf	66	1

LIS: library and information science.

The majority of the top 10 most frequently retweeted institutional users were iSchools members (@iSchoolsU, @UW_iSchool, @umsi, @SISCSU, @InfoSchoolSheff, @iusoic and @gslis) or SLA accounts ( @SLANewEngland and @SLAconf ). These users were retweeted an average of 126 times, usually by two organisations. @iSchoolsU, @umsi, @InfoSchoolSheff made both top 10 lists, demonstrating that, overall, iSchools users were most retweeted by these organisations.

The list of top 10 most commonly retweeted individual users was selected using similar analysis metrics. As summarised in Table 7, the most commonly retweeted individual user accounts include LIS educators, librarians, OCLC researchers and chief executive officers (CEOs) providing career advice and postings. These users were retweeted an average of seven times, usually by three organisations. The top 10 most frequently retweeted individual users list included librarians, the Director of Pew Research Center and executive leadership from ALA, SLA and ALISE. They were retweeted an average of 26 times, but usually by only one organisation.

Table 7.

Top 10 most retweeted individual users

Metric	Username	Profile^a	Frequency	Number of LIS associations
Most commonly retweeted	jill_hw	LIS Assistant Professor; a former member of the SLA Board of Directors	12	3
	Keribrary	Librarian; @ALCTS ED	9	3
	INALJNaomi	CEO, CMO #libraryjobs	8	3
	adamworrall4	LIS Assistant Professor	8	3
	Joanebeaudoin	LIS Assistant Professor	8	3
	Mmsubram	LIS Assistant Professor	7	3
	hassan_zamir	LIS researcher	6	3
	Aarontay	Librarian	5	3
	MerrileeIAm	OCLC Researcher	5	3
	Ernestopriego	LIS researcher, Lecturer	3	3
Most frequently retweeted	LibrarySherpa	2015 SLA Chair	110	1
	AnUncivilPhD	LIS Assistant Professor; Director of External Relations for the ALISE	33	1
	librarycourtney	2014–2015 ALA President	21	1
	sari_feldman	2015–2016 ALA president	16	1
	Coplibrarian	Instructor of library services	16	1
	Lrainie	Director of Pew Research Center	14	1
	Ljbphd	ALISE Board of Directors	14	2
	Stevenmaccall	LIS Assistant Professor	13	2
	MizCaramelVixen	Blogger; Writer	13	1
	Tranlib	Transportation librarian	13	1

LIS: library and information science; SLA: Special Libraries Association; ALA: American Library Association; ALISE: Association for Library and Information Science Education.

aThe profiles were accurate at the time of data collection and may change over time.

It was noted that there was no overlapping between the two top 10 lists in Tables 6 and 7, and no individual tweet was retweeted more than three times by over three organisations. In comparison, the top 10 most retweeted institutional users were retweeted significantly more often and by more organisations. This result suggests that user tweet influence is mainly among institutional users in LIS and the tweet influence of individual users is limited.

4.4.2. User mentions by the five LIS professional organisations

As shown in Table 8, only a mere six users received mention by all five organisations, accounting for 0.2% of the 3329 unique users mentioned. The vast majority of unique user mentions, 2978, were by only one organisation, accounting for 89.45% of those mentioned. Only about 10% of these users were mentioned by more than one organisation, each of which also mentioned their own special user groups.

Table 8.

Top 10 most mentioned institutional users

Metric	Username	Frequency	Number of LIS organisations
Most commonly mentioned	SchoolsU	56	5
	gslis	29	5
	nytimes	27	5
	CILIPinfo	15	5
	ProQuest	10	5
	WIRED	10	5
	nypl	37	4
	LibraryJournal	17	4
	US_IMLS	16	4
	TheAtlantic	15	4
Most frequently mentioned	JASIS&T	305	1
	SLAconf	197	1
	YouTube	160	3
	umsi	93	3
	iconf	77	3
	SLANewEngland	72	1
	SJSUASIS&T	65	2
	iSchoolsU	56	5
	SLAeurope	50	2
	SLALMD	49	1

LIS: library and information science.

Using similar analysis metrics, as described above, the top 10 most mentioned institutional users were summarised in Table 8. The top 10 most commonly mentioned institutional user list shows a diversified field distribution. Besides users from iSchools, others commonly mentioned were LIS organisations, public libraries, LIS magazines, vendors and notable media accounts (@nytimes, @WIRED, @TheAtlantic). Among these, the average number of times being mentioned was 23, by generally 4.6 organisations. The top 10 most frequently mentioned institutional users include branches of ASIS&T, SLA and members of iSchools. Among these users, the average number of mentions was 112, by two organisations on average, but only @ISchoolsU appeared on both lists.

As shown in Table 9, the top 10 most commonly mentioned individual users list features such diversified profiles as LIS instructors, taxonomists, the director of the Digital Public Library of America (DPLA), writers, editors and executive leadership of ALA, SLA and ALISE. These users averaged 15 mentions, usually by three organisations. The top 10 most frequently mentioned individual users were mentioned an average of 28 times by at least two organisations. The users @librarycourtney , @Musebrarian and @mmkhlava made both lists but were mentioned less frequently and by fewer organisations than the most mentioned institutional users.

Table 9.

Top 10 most mentioned individual users

Metric	Username	Profile^a	Frequency	Number of LIS organisations
Most commonly mentioned	librarycourtney	2014–2015 ALA President	58	5
	Musebrarian	LIS Assistant Professor	27	3
	mmkhlava	Taxonomist	16	3
	needalibraryjob	Unclear	12	3
	barbstripling	LIS Assistant Professor; President of ALA	8	3
	edrabinski	Coordinator of Library Instruction; editor	8	3
	rdlankes	LIS Professor	8	3
	jill_hw	LIS Assistant Professor; a former member of the SLA Board of Directors	7	3
	dancohen	Executive Director of DPLA	5	3
	doctorow	Writer, blogger, activist.	5	3
Most frequently mentioned	LibrarySherpa	SLA Chair of the 2015	59	2
	librarycourtney	2014–2015 ALA President	58	5
	Musebrarian	LIS Assistant Professor	27	3
	sari_feldman	2015–2016 ALA president	25	2
	leighgallagher	Fortune editor; host	23	1
	neilhimself	Novelist and screenwriter	19	1
	SKHastings	Past president of ALISE and ASIS&T; LIS Professor	17	2
	Mmkhlava	Taxonomist	16	3
	chasingseals	Professor of learning technologies	16	1
	LizLieutenant	Masters of LIS candidate	15	1

The profiles were accurate at the time of data collection and may change over time.

4.5. Interactions among LIS organisations

Surprisingly, the five LIS organisations studied here rarely intersect, as the number of their mutual RT accounted for 0.8% (47/6206) of total RT, while mutual mentions accounted for only 1% (71/6878) of total mentions. Their preferred topics included conferences, webinars, award nominations, grants, jobs and tweets that mentioned themselves.

4.5.1. Mutual RT between the five LIS professional organisations

Table 10 shows mutual RT between the five LIS organisations. ALA was the most commonly retweeted organisation by the other four, although only 12 times in total, while iSchools was the organisation others retweeted least. ALISE was most frequently retweeted by three associations, although 17 of their 19 RT were by ASIS&T. ALISE retweeted the other associations most frequently, at 19 times, and ALA retweeted the others the least.

Table 10.

Mutual retweets between LIS organisations

Metric	ALA	SLA	ASIS&T	ALISE	iSchools	Total frequency of being retweeted
ALA (being retweeted by)	–	3	1	7	1	12
SLA (being retweeted by)	1	–	0	1	1	3
ASIS&T (being retweeted by)	0	1	–	9	1	11
ALISE (being retweeted by)	1	0	17	–	1	19
iSchools (being retweeted by)	0	0	0	2	–	2
Total frequency of retweeting	2	4	18	19	4

4.5.2. Mutual mentions between the five LIS professional organisations

Table 11 shows mutual mentions between the five LIS professional organisations. Similar to the frequency of being retweeted, ALA was the most frequently mentioned by the other four LIS associations for a total of 27 mentions, while iSchools had the fewest mentions by the others. ALISE mentioned others most frequently (33 times), followed by ASIS&T (30 times).

Table 11.

Mutual mentions between LIS organisations

Metric	ALA	SLA	ASIS&T	ALISE	iSchools	Total frequency of being mentioned
ALA (being mentioned by)	–	2	5	16	4	27
SLA (being mentioned by)	1	–	8	1	0	10
ASIS&T (being mentioned by)	0	0	–	12	1	13
ALISE (being mentioned by)	0	0	14	–	0	14
iSchools (being mentioned by)	0	0	3	4	–	7
Total frequency of mentioning	1	2	30	33	5

4.6. Network analysis

The usernames of those accounts retweeted and mentioned by the five professional organisations were extracted from the total number of tweets and, respectively, input into the Gephi programme using spreadsheets in CSV format. Then, the RT network (R) and the mentions network (M) were constructed. Gephi represented Twitter user accounts as ‘nodes’, the communication paths between two nodes as an ‘edge’ and the width of edges are proportional to the total number of tweets recorded between two nodes. Researchers explored the status of interactions and connections among these organisations by three metrics of network: topological analysis, centrality analysis and community analysis [31 –33].

4.6.1. Topological analysis

A RT network (R network) including 1866 nodes and 2114 edges was constructed. In this graph, the nodes were those users (the five organisations) who either sent out or received ‘RT @username’ tweets. The edges were the relationships established between them by those tweets.

The mentions network (M network) was constructed including 3329 nodes and 3803 edges. The nodes were those users of the five organisations who either sent or received tweets with ‘@username’ mentions. The edges, again, were the relationship resulting from those tweets. The network diameter was found to be four in the R network and three in the M network, the longest path between two nodes in the network. The average path length was 2.008 in the R network and was 1.758 in the M network, indicating all these users were separated by about two nodes on average. To a certain extent, network diameter and average path length reflected the extent, ease and speed of information transmission, with the above data results indicating these organisations disseminated information along a shorter path and at faster speeds.

4.6.2. Centrality analysis

Degree centrality is the number of relationships that connect to a specific node. The higher the degree centrality, the more actors the node has contact within the network, thus the greater effect the node exerts in the network. Betweenness centrality is a measure of a node’s structural position, assessing the importance of nodes in the network. A node with high betweenness centrality tends to have a large influence on information transfer through the network, under the assumption that information transfer follows the shortest paths [31,33,34]. These two centrality measures of the five LIS professional organisations are shown in Table 12 and explained below:

ALA had the highest degree centrality in R network, and iSchools had the highest in M network, suggesting that both exerted more power in the networks.

ALA had the highest betweenness centrality in both networks, suggesting that it took the central position and had a great effect on information flow.

Interestingly, ALISE had a relative high betweenness centrality (2655) in the R network, but the lowest (755) in the M network, indicating that it was mostly a bridging mechanism in the R network but held only a marginal role in the M network.

SLA had relative low centrality in both networks, demonstrating its marginal role in LIS organisation tweets.

Table 12.

Centrality of five LIS professional organisations’ Twitter accounts

Metric	Degree centrality		Betweenness centrality
	R	M	R	M
ALA	637	847	2816	3264
SLA	329	464	1309	1660
ASIS&T	319	959	1336	2193
ALISE	411	574	2655	755
iSchools	433	976	1372	1632

4.6.3. Community analysis

Modularity measures the strength of division of a network into modules (also called groups, clusters or communities). Networks with high modularity have dense connections between the nodes within modules but sparse connections between nodes in different modules. Several clusters were identified by the modularity clustering tool embedded in Gephi. Modularity of the networks were 0.688 (R) and 0.696 (M), and five major sub-networks were found. Table 13 summarises the modularity class measures. In the R network, the ALA was the largest community, represented by 31.94% of the total nodes. The smallest community was ASIS&T, represented by 14.36% of the total nodes. In the M network, the largest community was ASIS&T and iSchools, which represented 24.63% of the total nodes, and SLA was the smallest, represented by 12.65% of the total nodes. The network diagrams for RT (R) and mentions (M) are illustrated in Figure 2. The colour of the nodes represents their modularity class. The largest (red) class is ALAlibrary in R networks, which consists of the users retweeted by ALALibrary. The smallest (purple) class in M network is ALISE, indicating ALISE mentioned fewest users in Twitter. Five main modularity classes have relatively clear boundaries, indicating that these five professional organisations have their own distinct retweeting or mentioning user groups.

Table 13.

Modularity of five major sub-networks

Metric	Modularity class
	R	M
ALA	31.94%	23.79%
SLA	16.61%	12.65%
ASIS&T	14.36%	24.63%
ALISE	17.20%	14.30%
iSchools	19.88%	24.63%

ALA: American Library Association; SLA: Special Libraries Association; ASIS&T: Association for Information Science and Technology; ALISE: Association for Library and Information Science Education.

Figure 2.

Network diagram for retweets (R) and network diagram for mentions (M)

Clustering coefficient is a measure of the degree to which nodes in a graph tend to cluster together and can be an indicator of network correlation characteristics. The community analysis showed a very low-average clustering coefficient (0.08 in R and 0.071 in M), indicating that the entire network was loosely and widely distributed (not cohesive). In order to explore the correlation characteristics among the five organisations, using the Gephi filters function by modularity class, the average clustering coefficient of the different subnets composed by the five major sub-networks (as composed by every two subnets, by every three, by every four) was analysed, with highest and lowest selected. Table 14 shows the results. The subnets composed of ALA and SLA in the R network and ASIS&T and SLA in the M network had average clustering coefficients much lower than the entire network. Further investigation revealed that the average clustering coefficient of the subnet made of SLA with other organisations was much lower, indicating that they had lower correlation with the other organisations. The subnet made of ALISE, iSchools and ASIS&T in both the R and M networks had a much higher average clustering coefficient than the entire network, demonstrating that these organisations maintained relatively close contact.

Table 14.

Average clustering coefficient of several subnets

Metric	Observatory network (R)	Average clustering coefficient	Observatory network (M)	Average clustering coefficient
Entire network	Retweets	0.08	Mentions	0.071
Pairwise comparison	Subnet composed by ALA and SLA	0.008	Subnet composed by ASIS&T and SLA	0.007
Pairwise comparison	Subnet composed by ASIS&T and ALISE	0.08	Subnet composed by ASIS&T and ALISE	0.07
Tri-wise comparison	Subnet composed by ALA, SLA and iSchools	0.015	Subnet composed by ALA, SLA and ASIS&T	0.017
Tri-wise comparison	Subnet composed by ASIS&T, ALISE and iSchools	0.093	Subnet composed by ASIS&T, ALISE and iSchools	0.085
Qualtra-wise comparison	Subnet composed by ALA, SLA, iSchools and ASIS&T	0.027	Subnet composed by ALA, SLA, ALISE and iSchools	0.038
Qualtra-wise comparison	Subnet composed by ALA, ASIS&T, ALISE and iSchools	0.084	Subnet composed by ALA, ASIS&T, ALISE and iSchools	0.077

ALA: American Library Association; SLA: Special Libraries Association; ASIS&T: Association for Information Science and Technology; ALISE: Association for Library and Information Science Education.

5. Discussion

5.1. What are the characteristics of LIS association tweets?

Twitter users have different methods to elicit communication and disseminate information: publicising news and events with hashtags, engaging others through direct user mentions or RT and using URLs to spread information and mobilise advocacy groups [35].

The results of this study show that LIS organisations’ tweets appear more conversational and engaging when compared with public or general scholarly tweets. For example, a study of random, public tweets revealed the rate of retweeting was just 3%, while 22% of tweets contained URLs and 36% mentioned users [35]. Another public timeline data set showed that only 12% of tweets contained conversations, while just 13% contained URLs [36]. These rates are significantly lower than the five organisations studied here, where there was a 40% retweet rate, with 44% of tweets mentioning users and 82% containing URLs. Previous research has suggested that there are disciplinary differences in scholarly tweet characteristics. For example, biochemists, at 42%, retweet substantially more than researchers in other disciplines; digital humanities and cognitive science researchers used Twitter more conversationally (38%); and economics researchers shared the most links (38%) [37].

The rate of LIS organisations’ tweets that contain hashtags also draws attention. Hashtags usually indicate topical tweets, tagged for users with similar interests to follow and find. In total, 55% of these tweets contained at least one. In contrast, in a sampling of tweets by political users in Singapore, only a very small fraction of their original tweets contained hashtags (<8%) [38]. In scientific contexts, Haustein et al. [39] analysed the tweeting behaviours of 37 astrophysicists. They found that almost one-quarter of examined tweets contained at least one hashtag (23.4%). It was worth noting that these characteristics of tweeting differences may vary from different approaches to collecting Twitter data, different account types or different data set sizes.

5.2. What topics or concerns do these LIS association tweets share?

Using text mining analysis by NVivo, combined with manual categorisation, seven major categories of tweets were identified (Libraries (all types) & services, Research, Conferences/Webinars/Continuing Education, Information Concepts, LIS education, Librarians and Jobs, Social Media). The content categorisation here differed greatly from academic and public library-specific categorisations [40,41], which are more detailed regarding various user- and collection-centered services and activities. The content categories of LIS professional associations’ tweets are broad in nature, giving an overarching view of development of practice and theory. This allows LIS professionals to learn from these associations by following current topics of conversation, through networking. While each association had its own preferences, all attached great importance to conference hashtags.

The research findings showed that LIS association topics were very diverse, as evident by the number of unique hashtags (2354) in the sample data. The distribution of these hashtags showed that very few receive wide use. The most popular and frequent topics, by all five organisations, included #library, #archives, #librarians, #bigdata, #socialmedia and #lis. In addition, LIS associations paid close attention to topics involving user-motivated interactions and engagement across boundaries. For example, SLA used #SLAtalk and ASIS&T launched #10MinReads. Extensively used, both hashtags provide different perspectives on common, wide-ranging issues.

5.3. Who are the Twitter users involved in the LIS associations’ tweets?

Social media use is voluntary, so it is important that an organisation attracts a critical mass of followers and facilitates active participation in their online communities [42]. RT or mentions may be important methods for attracting users [43]. LIS associations made full use of ‘word-of-mouth’ interactions between third-party stakeholders, retweeting and mentioning many of these users.

A total of 1866 unique retweeted users were found in the data set, which was far fewer than the number of users mentioned (3329). The results of user analysis showed that only a very few popular users were commonly retweeted or mentioned by LIS professional organisations, and most users were infrequently retweeted or mentioned.

This study’s user analysis revealed the visible users in these five organisations’ tweets. The top 10 most retweeted users mainly represented the LIS field, including the leadership and branches of other professional organisations and researchers. The top 10 most mentioned users showed more diversified distribution, including news media, vendors and writers. Institutional users tended to receive more attention than individuals, receiving more RT and mentions by these organisations. The growing importance of iSchools is reflected in their members’ large percentage on both of these top 10 user lists.

5.4. What are the interactions among the LIS associations on Twitter?

One of the goals of this study was to explore how LIS professional organisations interact with each other on Twitter. Community analysis showed a very low-average clustering coefficient in networks of RT and mentions, indicating that these organisations seldom interact and that it is a loosely connected network of LIS professional organisations overall on Twitter. However, the average clustering coefficient of the subgroups consisting of ALISE, iSchools and ASIS&T were relatively higher than others, suggesting that this subset of LIS associations is closely connected.

Betweenness centrality analysis showed that these five professional organisations played different roles in information diffusion on Twitter. ALA was the most influential and played a role as a central node, possibly due to its many divisions and initiatives and well-established communication channels. The betweenness centrality of SLA was much lower, showing its marginal role in the network. ALISE had the more influential nodes in the RT network, but not in the mentions network, which is the opposite of ASIS&T. As explained by Cha et al. [29] that RT were content-driven and mentions more conversation-driven, the result may mean that ALISE was more focused on LIS content, while ASIS&T was more interested in facilitating user conversations.

5.5. Practical implications

The findings of this study have several practical implications to LIS professional development, institutional use of Twitter, LIS research and LIS professional organisations.

5.5.1. Professional development

LIS associations’ tweets are both informative and conversational in nature and serve as a valuable information source for professional development. In particular, these tweets contain unique topics, broadly ranging from educational opportunities and conferences, to self-paced instructional materials for individualised learning and networking. Learning is critical for the success of all professionals, which is particularly true for LIS professionals in an ever-changing information and technology landscape. LIS association Twitter accounts are important platforms for conveying the latest news and events, acquiring new knowledge and skills and raising awareness of issues related to the LIS field. In addition, the selectively retweeted tweets of popular and reputable professional experts could enable new professionals to rapidly build professional networks in the Twitter space.

5.5.2. Institutional use of Twitter

LIS associations’ tweets also broadcast employment opportunities, market new products and services and display risk management methods from which institutions could benefit. LIS associations’ tweets predominately consist of librarian- and job-related information. LIS institutions can take advantage of the large and relevant user base for recruitment purposes and identify and screen potential candidates or experts through analysing their user profiles and timelines (e.g. tweets, RT and followings). LIS associations’ tweets tend to include new products and services such as 3D Printing, digital preservation and curation and Linked Data, with their followers offering feedback and reviews. Libraries or other institutions could follow developing trends or promote and market their own products and services to enhance their reputations and raise visibility within the LIS community. Additionally, LIS associations’ Twitter accounts serve as an effective, real-time collaborative platform to broadcast and monitor ongoing events and assess developing problems in the LIS community.

5.5.3. LIS research

The variety of research topics covered in LIS associations’ tweets can be valuable for researchers to identify focus points and emerging LIS topics and issues that critically concern the LIS community. The LIS associations’ tweets follow and engage research activities, as shown in our themes’ analysis that research is the second largest discussion category. These topics include information regarding research activities, dissemination of research results, competitive awards helping to fund research projects and research forums and meetings. It is important that LIS researchers, professionals, instructors and students be keenly aware of the current main spheres of research within the discipline.

5.5.4. LIS professional organisations

The results of this study could help LIS professional organisations review and evaluate their own Twitter use strategies. Today’s LIS professional organisations rely on an active membership to develop and deliver their products and services. While they can take advantage of Twitter as an electronic word-of-mouth for attracting and retaining audiences and members, the current use has been mainly one-way communication from associations without much engagement and interactive activities with targeted audiences. They could better utilise Twitter’s interactive community-building features such as mentions, RT and hashtags to develop, engage, retain and sustain user communities on Twitter.

The comparative study and the patterns identified among LIS associations’ tweets are helpful for understanding each organisation’s unique position in the LIS Twitter space, which is extremely valuable for outreach to intended user audiences and more effective social media use. LIS professional organisations can tailor their Twitter outreach activities to targeted audiences.

6. Conclusion and future research

While LIS associations have been utilising Twitter, common and best practices are still evolving and little understood or researched. This study addresses this gap through an analysis and comparisons of Twitter utilisation by five major US-based LIS associations, through a systematic framework: descriptive analytics, content analytics, user analysis and network analytics. The study contributes to an efficient approach for analysing the content of LIS tweets through combining automated text mining with manual categorisation, in contrast to the broad information technology literature focused on automated analyses, and LIS literature mainly involving manually extracting and analysing data from small samples of tweets.

The results of this study reveal the basic characteristics of the Twitter usage, topic discussions, user types, user influences, organisation interaction patterns and the role these organisations played in communicating and distributing information within Twitter for LIS and related fields. Further research could enhance our understanding of social media use in LIS professional organisation contexts, with an extended sample in a longer time frame, focusing on more detailed and systematic analysis. Future research may utilise network analysis with co-hashtag graphs and hashtag-user graphs to discover whether information content among user groups differs, develop in-depth measurements of user influence combined with more metrics, investigate user influence on diverse topics and explore the spatiotemporal analyses of Twitter data.

Footnotes

Acknowledgements

The authors would like to thank Donald Jay Kulpa and Wendy Bromfield for their research assistance for this project. This paper won the 2017ALISE/Bohdan S. Wynar Research Paper Competition.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.

Funding

This study is supported in part by China Scholarship Council (CSC) (no. 201406995058) and by China National Social Science Research Fund for the project Research on the model of information aggregation and sharing in network academic community (no. 11CTQ038) received by the first author (M.Z.).

References

Carscaddon

Chapman

. Twitter as a marketing tool for libraries. In: Thomsett-Scott

(ed.) Marketing with social media: a LITA guide. Chicago, IL: American Library Association, 2013, pp. 1–21.

Milstein

. Twitter for libraries (and librarians). Comput Libr 2009; 29(5): 34–35.

Dickson

Holley

. Social networking in academic libraries: the possibilities and the concerns. New Libr World 2010; 111(11–12): 468–482.

Hsiang

. Effectiveness of social networking in a university library environment, http://gra103.aca.ntu.edu.tw/gdoc/99/D93922021a.pdf (2010, accessed April 2016).

Noordegraaf

. Remaking professionals? How associations and professional education connect professionalism and organizations. Curr Soc 2011; 59(4): 465–488.

Henczel

. The impact of library associations: preliminary findings of a qualitative study. Perform Measure Metric 2014; 15(3): 122–144.

Ghosh

. The emerging role of national and regional associations in library development: an Indian perspective. Libr Rev 2006; 55(1): 45–58.

Madden

. Today’s UK professional association library and information service challenges and best practice. ASLIB Proc 2008; 60(6): 556–569.

Friedman

Williams

. Member relations and strategy: supporting member involvement and retention. Bristol: Professional Association Research Network, 2006; pp. 1–155.

10.

Sykes

. Information needs assessment. In: Scammell

(ed.) Handbook of information management. 8th ed. London: Aslib-IMI, 2001, pp. 114–131.

11.

Soules

. Forecasting the future. In: Sheehy

(ed.) Managing business collections in libraries. Westport, CT: Greenwood Press, 1996, pp. 217–234.

12.

Robinson

. Marketing preparation: the key to success. Lib Info Update 2006; 5(3): 32–33.

13.

Spiller

Creaser

Murphy

. Libraries in the workplace. Loughborough: Library and Information Statistics University, 1998.

14.

Orna

. Information strategy in practice. Aldershot: Gower Publishing, 2004.

15.

Singh

Gill

. Web 2.0 technologies in libraries: a survey of periodical literature published by Emerald. Lib Rev 2013; 62(3): 177–198.

16.

Hussain

. Adoption of Web 2.0 in library associations in the presence of social media. Program 2015; 49(2): 151–169.

17.

Phillips

. Facebooking it: promoting library services to young adults through social media. Publ Libr Q 2015; 34(2): 178–197.

18.

Romero

. ROI. Measuring the social media return on investment in a library. Bottom Line 2011; 24(2): 145–151.

19.

Vassilakaki

Garoufallou

. The impact of Twitter on libraries: a critical review of the literature. Electron Libr 2015; 33(4): 795–809.

20.

Zohoorian-Fooladi

Abrizah

. Academic librarians and their social media presence: a story of motivations and deterrents. Inform Dev 2014; 30(2): 159–171.

21.

Vassilakaki

Garoufallou

. The impact of Facebook on libraries and librarians: a review of the literature. Program 2013; 48(3): 226–245.

22.

Sewell

. Who is following us? Data mining a library’s Twitter followers. Libr Hi Tech 2013; 31(1): 160–170.

23.

Shulman

Yep

Tomé

. Leveraging the power of a Twitter network for library promotion. J Acad Libr 2015; 41(2): 178–185.

24.

Burclaff

Johnson

. Developing a social media strategy: tweets, pins and posts with a purpose. Coll Res Libr News 2014; 75(7): 366–369.

25.

Kim

Hansen

Murphy

. Methodological considerations in analyzing Twitter data. J Natl Cancer I Monograph 2013; 47: 140–146.

26.

Chae

B(K)

. Insights from hashtag #supplychain and Twitter analytics: considering Twitter and Twitter data for supply chain practice and research. Int J Prod Econ 2015; 165: 247–259.

27.

Zimmer

Proferes

. A topology of Twitter research: disciplines, methods, and ethics. Aslib J Inform Manage 2014; 66(3): 250–261.

28.

Fisher

. The value of professional association. Libr Trends 1997; 46(2): 320–330.

29.

Cha

Haddadi

Benevenuto

. Measuring user influence in Twitter: the million follower fallacy. In: Proceedings of the fourth international AAAI conference on weblogs and social media, Washington, DC, 23–26 May 2010, http://snap.stanford.edu/class/cs224w-readings/cha10influence.pdf (accessed April 2016).

30.

Bruns

Stieglitz

. Towards more systematic Twitter analysis: metrics for tweeting activities. Int J Soc Res Methodol 2013; 16(2): 91–108.

31.

Maruyama

Suthers

Robertson

. Characterizing communication networks associated with political hashtags. In: Proceedings of 47th Hawaii international conference on system science (HICSS), Waikoloa, 6–9 January 2014.

32.

Palmer

. Characterizing university library use of social media: a case study of Twitter and Facebook from Australia. J Acad Libr, 2014; 40(6): 611–619.

33.

Raamkumar

Pang

Foo

. When countries become the talking point in microblogs: study on country hashtags in Twitter. First Monday 2016; 21(1). http://journals.uic.edu/ojs/index.php/fm/article/view/6101/5193

34.

Luo

Zhong

. Using social network analysis to explain communication characteristics of travel-related electronic word-of-mouth on social networking sites. Tourism Manage 2015; 46: 274–282.

35.

Boyd

Golder

Lotan

. Tweet, tweet, retweet: conversational aspects of retweeting on Twitter. In: Proceedings of the 43rd Hawaii international conference on system sciences IEEE. Washington, DC: IEEE Computer Society, 2010, pp. 1–10.

36.

Java

Song

Finin

. Why we Twitter: understanding microblogging usage and communities. In: Proceedings of the 9th WebKDD and 1st SNA-KDD 2007 workshop on web mining and social network analysis. New York, NY: ACM, 2007, pp. 56–65.

37.

Holmberg

Thelwall

. Disciplinary differences in Twitter scholarly communication. Scientometrics 2014; 101(2): 1027–1042.

38.

Kywe

Hoang

Lim

. On recommending hashtags in Twitter networks. In: Aberer

Flache

Jager

. (eds) Social informatics. Berlin: Springer–Verlag, 2012, pp. 337–350.

39.

Haustein

Bowman

Holmberg

. Astrophysicists on Twitter: an in-depth analysis of tweeting and scientific publication behavior. Aslib J Inform Manage 2014; 66(3): 279–296.

40.

Aharony

. Twitter use in libraries: an exploratory analysis. J Web Librarian 2010; 4(4): 333–350.

41.

Shiri

Rathi

. Twitter content categorisation: a public library perspective. J Inform Knowl Manage 2013; 12(4): 1350035.

42.

Culnan

McHugh

Zubillaga

. How large U.S. companies can use Twitter and other social media to gain business value. MIS Quart Execut 2010; 9(4): 243–259.

43.

Palmer

. Characterisation of the use of Twitter by Australian universities. J Higher Educ Policy Manage 2013; 35(4): 333–344.