Quality Management in Social Business Intelligence Projects

María José Aramburu

, Rafael Berlanga

and Indira Lanza-Cruz

Department de’Enginyeria i Ciència dels Computadors, Universitat Jaume I, E-12071 Castelló de la Plana, Spain

Department de Llenguatges i Sistemes Informàtics, Universitat Jaume I, E-12071 Castelló de la Plana, Spain

Keywords: Quality Management, Social Data, Business Intelligence.

Abstract: Social networks have become a new source of useful information for companies. Increasing the value of social

data requires, first, assessing and improving the quality of the relevant data and, subsequently, developing

practical solutions that apply them in business intelligence tasks. This paper focuses on the Twitter social

network and the processing of social data for business intelligence projects. With this purpose, the paper starts

by defining the special requirements of the analysis cubes of a Social Business Intelligence (SoBI) project

and by reviewing previous work to demonstrate the lack of valid approaches to this problem. Afterwards, we

present a new data processing method for SoBI projects whose main contribution is a phase of data exploration

and profiling that serves to build a quality data collection with respect to the analysis objectives of the project.

1 INTRODUCTION

Social networks have become a new source of useful

information for companies, helping them, among

others, to know the opinions of their customers, to

analyse the trends of the market, and to discover new

business opportunities (Ruhi, 2014). Although social

media data are highly heterogeneous and difficult to

manage, they can produce meaningful information for

decision-makers. The research here presented focuses

on how to build quality social media data collections

for Social Business Intelligence (SoBI) projects.

SoBI is defined in (Gallinucci et al., 2015) as the

discipline that aims at combining corporate with

social media data to let decision-makers analyse and

improve their business needs based in the trends and

moods perceived from de environment. Until now,

companies have used social networks mainly for

marketing purposes. SoBI tools are often applied by

marketing departments to monitor the results of their

activities by means of a group of social media metrics

(e.g. number of likes, followers, or replies) (Keegan

& Rowley, 2017) and methods (Lee, 2018). However,

social media metrics are rarely combined with other

business measures to calculate key performance

indicators of different purpose (Ruhi, 2014). The

integration of social media metrics with corporate

data to produce new strategic indicators are some of

the SoBI related applications that can bring new

opportunities to companies (Agostino et al., 2018).

In current SoBI projects, filtering Twitter by

means of a set of keywords (i.e., topics and hashtags)

generates a flow of potentially relevant tweets. By

processing them, a group of measures and attributes

is extracted which are used to calculate metrics and

indicators. However, most times, in the large volume

of data retrieved, there are many tweets apparently

related to the subject of analysis but that, because of

their origin, intention or specific contents, are not

useful. Therefore, we consider that before processing

the tweets, some filtering tasks are necessary.

However, previous frameworks for social media

analytics have not given the required importance to

data quality in their data preparation phases (Stieglitz

et al., 2018).

In this paper, we highlight the importance of

building a quality data collection in the data

preparation phase of SoBI projects. After defining in

Section 2 the special requirements of the analysis

cubes of a SoBI project and reviewing, in Section 3,

previous work to demonstrate the lack of valid

approaches to this problem, we present, in Section 4,

a new data processing method for SoBI projects. Its

main contribution is a new phase of exploration and

profiling of the retrieved data that serves to build a

quality data collection with respect to the analysis

objectives of the project.

320

Aramburu, M., Berlanga, R. and Lanza-Cruz, I.

Quality Management in Social Business Intelligence Projects.

DOI: 10.5220/0010495703200327

In Proceedings of the 23rd International Conference on Enterprise Information Systems (ICEIS 2021) - Volume 1, pages 320-327

ISBN: 978-989-758-509-8

2 BUSINESS INTELLIGENCE

WITH SOCIAL MEDIA DATA

Traditional Business Intelligence (BI) projects apply

several types of tools to exploit the facts of a subject

of analysis, like for example, the sales of a company.

Usually, these facts are represented into the analysis

cubes of a multidimensional data model. A subject of

analysis is a business context dependent description

that clearly determines the analysis objectives of the

project, and consequently, the measures and

dimensions of its analysis cubes.

In the case of a SoBI project, the facts that serve

to feed the analysis cubes come from a collection of

user-generated contents (e.g. tweets). These contents

are external to the company and consists of

unstructured data with a high level of heterogeneity,

which makes much more difficult to obtain valid

facts. Here, the main issue is that by simply

translating the subject of analysis into a group of

keywords, the retrieved tweets will include many

posts generated with very different purposes and out

of the scope of the SoBI project. Consequently, many

of the retrieved posts do not add any value to the

analysis tasks and may even be counterproductive,

due to the misinformation and the noise that they can

introduce in the intended analysis. For example, in

our experiments, when retrieving tweets with

opinions from the users of a Ford car model, it is

impossible to avoid the retrieval of many memes

about the actor Harrison Ford and the words “fiesta”,

“escort” or “focus”, which also refers to different

Ford car models.

2.1 Defining Analysis Cubes

The subject of analysis of a SoBI project must be

contextualised in terms of the organization's strategic

objectives (Ruhi, 2014) (Lee, 2018). In our approach,

this task consists in defining the analysis objectives

of the project by means of an analysis cube, that is, of

a group of measures which will be analysed from

different points of view and levels of detail. Among

these measures, there can be metrics extracted by

processing tweet contents together with other coming

from tweets metadata (e.g. number of followers,

replies, likes, …). The relevant data will depend on

the analysis objectives to be achieved. These are some

examples:

 If the objective is brand awareness, metrics

such as share of voice, are an indicator of the

visibility of the brand with respect to its

competitors.

 To gain engagement from customers or

prospects, audience metrics such as number of

comments, shares and trackbacks serve to

define good indicators.

 To improve customer service, monitor the

comments and negative sentiment of customers

to generate the right responses and solutions, as

well as indicators such as resolution time and

satisfaction scores after giving support.

Additionally, the dimensions of the analysis

cubes should represent all the attributes, categories,

and hierarchies that the analysis objectives of the

project require. For example, when the managers of a

car brand want to analyse the sentiments and opinions

about their models, they can apply several measures

and points of view. By processing tweet contents, it

can be possible to estimate measures such as the

general polarity of a tweet from opinions about both

the brand or the car model, as well as, the specific

aspects assessed by the user (i.e. engine, ecology,

design, …). The temporal dimension can be applied

to follow up the evolution of these measures, and the

spatial dimension would allow studies by considering

the geographical distribution of the sources of

opinion. However, to discover new valuable insights,

it would be interesting to have additional dimensions

for the categories of car models of the brand as well

as for the different car aspects. Furthermore, to have

the measures mapped against a dimension of types of

users would allow the execution of analysis

operations adapted to the different actors of this

application (e.g. journalists, professional drivers,

general drivers, …). Modelling all these elements

requires understanding the brand business model and

its strategic objectives.

When building an analysis cube, a main issue is

how to obtain a social media data collection from

where to extract the measures of the analysis cubes

along with all the dimensional attributes. To produce

reliable insights, the analysis cubes must be complete

and uniformly filled, with no gaps nor biased

dimensions. For this reason, it will be necessary to

retrieve enough tweets by means of a full range of

possible keywords and hashtags but, at the same time,

to avoid overloading the collection with redundant

and non-relevant tweets. Therefore, it is necessary to

execute filtering operations on the collection.

Determining the feasibility of the project and the

range of retrieval and cleansing operations needed to

build a quality social media data collection is a

complex task that current methodologies do not give

enough importance to.

Quality Management in Social Business Intelligence Projects

321

3 RELATED WORK

Defining a methodology to build quality collections

of social media data involves several things. By one

hand, it is important including in the methodology

some data cleaning operations. On the other hand, it

is necessary to determine the best quality metrics to

be applied by cleaning operations as well as to assess

the quality of the overall collection. In this section,

we review previous methodologies from the point of

view of quality management, and then, we summarise

the different approaches to measuring credibility, the

most important quality attribute for social media.

3.1 Previous Methodologies

Modern methodologies for processing Big Data

consider that quality management should take part of

all the phases of the pipeline (Taleb et al., 2018)

(Pääkkönen & Jokitulppo, 2017). Considering the

data acquisition phase, the Big Data quality

framework presented by (Taleb et al., 2015)

highlights the need for cleansing, integration, filtering

and normalization operations to improve the quality

of the collection and, in this way, to save on costs and

perform accurate data analysis.

The uniform data management approach of

(Goonetilleke et al., 2014) reviews three main groups

of research challenges to address when building a

Twitter data analytics platform. For data collection,

the main issue is the specification of the best set of

retrieval keywords and hashtags; for data pre-

processing, they demand for specific data processing

and extraction strategies for Twitter data; and finally,

for data management, they explain that quality

management is a major issue that requires declarative

languages to query social networks. However, as

noted by (Stieglitz et al., 2018), in the papers that

already document the data tracking and preparation

steps of their social media projects, these steps are

often dealt with superficiality, never with as much

extension as data analysis tasks. The authors conclude

that data discovery, collection and preparation phases

for social media analytics require more research.

In the quality management architecture for social

media data of (Pääkkönen, P. & Jokitulppo, J., 2017),

the data acquisition, data processing & analysis, and

decision-making phases can include functionalities

for quality control and monitoring. In this approach,

data quality management consists in assigning values

to quality attributes which can be applied at

extraction, processing, and analysis time from the

point of view of the data source, the data and the user

respectively. The quality, organizational and

decision-making policies of the organization define

the criteria to filter the quality data. Although, the

proposed architecture can represent all these data

quality elements, the authors do not propose a

methodology for defining and applying them.

The methodology for SoBI of (Francia et al.,

2016) recognizes that crawling design can be one of

the most complex and time-consuming activities and

aims at retrieving in-topic clips by filtering off-topic

clips. They also explain that filtering off-topic clips at

crawling time could be difficult due to the limitations

of the crawling languages and propose to filter them

at a later stage by using the search features of a

documents database. The authors note that manually

labelling a sample of the retrieved clips enables the

team to trigger a new iteration where the crawling

queries are redefined to cut off-topic clips out more

effectively. However, the proposed methodology

does not include any cleaning operations in its

functional design nor determines whether to process

them before or after retrieving the social data. This

work does not consider the quality of data as a main

objective, and it does not explain how to obtain a

good quality collection.

A second methodology for SoBI, proposed by

(Abu Salih, 2015), consists of five stages that process

social media data and integrate them in the data

warehouse files. They propose to execute cleaning

operations to remove dirty data at the data acquisition

stage prior to data storage. Afterwards, during data

analysis, the collected data is processed to infer a

value of trust for the relevant data. In the last stage of

this methodology, a data structuring process serves to

integrate traditional and social media data in order to

produce new insights. In this way, the exploited social

media data has a minimum level of trust with respect

to its domain, although, the analysis objectives of the

SoBI project are not considered as additional criteria

to validate the source data.

3.2 Quality Metrics for Social Media

Measuring social media data quality can be

performed using different metrics and techniques.

The literature review clearly reveals that credibility is

the most important quality attribute for social media,

and many different approaches have been proposed to

measure it. It is important to clarify that for these

authors, credibility is a broad concept that intersects

with other semantically related quality attributes such

as trust, reliability, believability, veracity, relevance,

validity and, in some cases, even understandability

and reputation.

ICEIS 2021 - 23rd International Conference on Enterprise Information Systems

322

For measuring the credibility of social media data,

statistical or machine learning techniques are usually

adopted. Among the many metrics used to feed these

algorithms, there are those obtained by processing

tweets contents, mainly looking for textual properties,

writing styles, linguistic expressions, sentiments, and

additional elements like URL´s or pictures. Social

parameters extracted from tweets metadata about the

post and its poster are a second source of metrics.

Finally, there are a last group of metrics with

information about the behaviour and actions of the

users in the social network. Table 1 shows metrics

applied in the literature to measure credibility (Sikdar

et al., 2013) (Gupta et al., 2014) (Viviani & Pasi,

2016) (Păvăloaia et al., 2020) and which, in many

cases, could also be applied to assess other quality

attributes. The wide range of metrics applied shows

that, in each approach, credibility can be understood

in a different way, and that it is up to the user to

choose the best metrics taking into account the

context of each project, its domain and the

technologies applied.

In social media, a quality attribute of utmost

importance is reputation which can be also defined as

the authority of the poster. Previous work frequently

considers simple measures such as the number of

followers to calculate indicators of good reputation,

i.e. when a user is in many Twitter lists and has many

followers is because the contents that generates

satisfy many users. This approach ignores that user’s

interest can be diverse and evolve and change over

time. However, recently, more realistic approaches

have proposed to consider this quality attribute as a

time and domain-dependent parameter (Abu-Salih et

al., 2020).

Finally, in (Pasi et al., 2019), the adequacy and the

potentialities to describe the issue of the assessment

of the credibility of user-generated content in social

media as a multi-criteria decision-making problem

have been discussed. Their approach to determining

the credibility of online reviews considers features

connected to the contents, the information sources

and the relationships established in social media

platforms. These features are evaluated by the users

in terms of their impact on veracity. By considering

different aggregation schema for the partial

performance scores and their impact, the authors

calculate an overall score of veracity. With respect to

data-driven approaches based on Machine Learning

techniques, their approach makes the user more aware

of the choices that led to the proposed decision and

can make the considered problem less data-

dependent.

In this section, we have reviewed relevant

methodologies and techniques to build collections of

social media data for SoBI from the point of view of

quality management. The conclusion is that while

most approaches to social media analysis for decision

support apply different quality criteria during data

collection and preparation, it is not clear at this point

how to define a general-purpose quality conceptual

model for social media data. Previous work has

provided us with many different metrics with multiple

purposes that depend primarily on the application.

The experience demonstrates that, whatever

technology applied to decide the quality of data, a

good combination of different types of metrics use to

be part of the solution. However, there is no

systematic methodology for defining and applying

these metrics in order to build a reliable collection of

social data. In this case, the main question is how to

find the best metrics that can be applied to both

prepare the collection and measure its quality with

respect to the analysis objectives of a SoBI project,

i.e. the construction of an analysis cube with a set of

social measures that can be analysed from different

points of view and levels of detail.

Table 1: Sample of metrics to measure credibility in social media data found in the literature.

Tweet contents Post and poster metadata User behaviour

# Chars/words

# Punctuation symbols

# Pronouns

# Swear words

# Uppercases

# Emoticons

#URLs/images

# Hashtags

# Misspelled words

# Sentences

Average length of sentences

# Product mentions

# Product features mentioned

# Opinion sentences

Account age

Listed count

Status count

Favourites count

# Friends

# Followers

# Followings

Ratio of followers to friends

Mean text length in tweets

Mean hashtags in tweets

Mean # URLs/ mentions in tweets

Verified user

User image in user profile

Tweet geographical coordinates

# Retweets

# Tweets

# Tweets favorited

# Mentions

# Tweets are a reply/retweet

Mean time between tweets

# Likes received

# Directed tweets

# Users that propagate the user

# Users the user propagates from

# Tweets propagated by other users

# Users that converse with the user

Mean number of conversations

Average length of chain-like behaviour

Quality Management in Social Business Intelligence Projects

323

4 QUALITY MANAGEMENT IN A

SoBI PROJECT

Evaluating the quality of social media data for a SoBI

project requires the definition of the best quality

indicators for the source data (Immonen et al., 2015).

Tweets present many different aspects that would

serve to filter them; being Tweet contents and users’

attributes and actions, the main sources of quality

metrics. However, the selection of the best quality

metrics for a SoBI project is a complex task that

requires a deep understanding of the business context

and objectives of analysis, as well as the social media

data to be managed (Berlanga et al., 2019).

In data intensive applications, quality conditions

should serve as criteria to program the cleaning

operations and to measure the quality of the overall

collection. As explained in (Sadiq & Indulska, 2017),

traditional methods for managing data quality follow

a top-down approach: the analysis of user

requirements produces some quality rules that serve

to govern data, to assess data quality, and to execute

cleaning operations. This approach is suitable for

managing the quality of data generated internally by

an organization. However, when the organization

does not control the external processes that generate

the available data, as in the case of social media data,

quality assessment requires prior knowledge about

the data. In these cases, the data quality management

follows a bottom-up approach that starts with the

execution of some exploratory analysis and data

profiling tasks. These tasks help to find data quality

rules and requirements that will drive the data

collection process. In order to execute the preliminary

exploration of the available data, interactive,

statistical and data mining techniques can be applied

over collections of data. Some of the data mining

techniques that can help in these tasks are clustering,

classification, data modelling and data summarization

(Stieglitz et al., 2018).

Following this approach, in this section we

present a new method for Social Media data

processing for SoBI applications whose main

contribution is a first phase of exploration and

profiling of the retrieved data that serves to build a

quality data collection with respect to the analysis

objectives of the project.

4.1 A Data Processing Methodology

The analysis objectives of a SoBI project are to

analyse from various perspectives (i.e. dimensions

and categories of analysis) the social measures drawn

from a complete collection of relevant tweets. Often

these will be simple measures such as the number of

likes or followers, especially if the aim is to analyse

the success of a marketing campaign. For example, if

the objective is brand awareness, social measures

such as voice share are an indicator of the brand's

visibility in relation to its competitors that can be

calculated by counting hashtags and mentions.

However, there are other more difficult measures to

analyse, such as when analysing feelings about

different aspects of a product or service (e.g. a holiday

package or a car).

To deal with the completeness of the collection

and the relevance of the tweets in it, in this paper we

propose a new data processing methodology that

consists of three main phases: Collection

Construction, Data Preparation and Data Exploitation

(see Figure 1). The first phase is the construction of a

collection of tweets through an exploratory process

executed by the user and directed by the quality of the

recovered data. When a quality data collection is

ready, in the data preparation phase, the facts of the

analytical cubes are extracted from the posts and then

exploited in the last phase of the process.

As Figure 1 illustrates, during the Collection

Construction phase, the user executes some data

exploratory and profiling tasks to assess and improve

data coverage and data quality until obtaining a

quality collection that meets the project's analysis

objectives. More specifically, this phase consists of

two complementary and iterative tasks:

a) Evaluating the subject coverage of the

collection with respect to its topics and users.

Analysing the vocabulary of the collection will

help to know when to redefine the keywords

applied to program the Twitter API to obtain

either a more complete or precise collection.

Furthermore, profiling the range of users that

post the retrieved tweets along with their

metadata is important in determining the

measures and dimension attributes available to

be part of the analysis cubes, as for example,

the users’ demographic data present in their

descriptions.

b) Analysing and improving the quality of the

collection by filtering the posts of low quality

or out of the scope. Finding the best quality

metrics that help to clean a collection requires

exploring and profiling it in order to discover

its main characteristics and the sources of

noise. Then, cleaning operation can apply

different types of quality metrics extracted by

processing the tweets contents and metadata as

well as the user descriptions. For example,

ICEIS 2021 - 23rd International Conference on Enterprise Information Systems

324

Figure 1: The proposed social media data processing method for Social Business Intelligence.

from a collection of user opinions about cars, memes

should be removed because their low added value as

well as the ad posts sent by marketing departments. A

global analysis enhances the detection of long-term

quality problems, such as redundancy, bias and noise,

which are often difficult to detect from local analysis

(i.e., directly over the streamed data). Analysing a

long-term data stream also allows building robust

language models for a given domain, smoothing the

effects of punctual viral events (Lanza-Cruz et al.,

2018).

In the Collections Construction and Data

Preparation phases, the processing of tweets to extract

the measures and values that serve both to clean the

collection and to feed the analysis cubes, can be made

in different ways. Some values are directly available

in the tweets metadata, such as post-date and number

of followers of the user. Other values can be

calculated with a simple processing like counting

tweets over a period. Evaluating the grammatical

richness of a post is executed by a process that

calculates some textual measures (Gupta et al., 2014).

However, more sophisticated Natural Language

Processing techniques are required when analysing

the vocabulary of the whole collection (Berlanga et

al., 2019) or extracting the facets and polarity of

opinion expressions (García-Moya et al., 2011).

Other intelligent techniques like Entity Resolution

and Ontology Mappings, mainly applied for the

Semantic Enrichment and Dimensional Mapping

tasks of Figure 1, have shown useful to extract

dimensional attributes from social media data

(Berlanga et al., 2015) (Pereira et al., 2018).

Furthermore, the tweets in the collection can be

applied to train Machine Learning algorithms that

discover some useful properties of the tweets, as for

example their credibility and intention, and to classify

them into analysis categories. The main objective of

all these tasks is to produce a complete analysis cube

whose measures and dimension attributes can enable

reliable multidimensional studies.

Finally, in the Data Exploitation phase, the

analysis cubes constructed by processing the tweets

collection can be stored into the Corporate Data

Warehouse for future uses. OLAP applications, or

any other Business Intelligence or Data Mining tools,

can be applied to analyse and extract new insights

from these cubes. If necessary, the stored social

media facts can be combined with corporate data to

design new key performance indicators tied to the

strategic business objectives such as ROI and profit

margins. In this way, users can associate social media

actions with sales volume, revenue increases or

decreases and other relevant metrics and reveal new

insights.

The focus on integrating social media metrics and

internal business measures is important in our

approach. However, combining qualitative data from

social networks with the quantitative data hosted in

traditional BI systems may seem a difficult task. Note

that as pointed out by (Ruhi, 2014) this will require

the semantic integration of the data elements in

both kinds of external and internal data sources. From

a practical point of view, this integration can

be possible if the dimensional parameters applied

to construct the analysis cubes of social media and

Quality Management in Social Business Intelligence Projects

325

corporate data are compatible (i.e. mappable).

At this point, it is important to note that the

success of a SoBI project will mainly depend on the

completeness of the analysis cube that can be

constructed. This means that sometimes, it will be

difficult or even impossible to build the collection of

tweets to fill the analysis cube in all its dimensions.

To this end, the data exploration and profiling tasks

of the Collection Construction phase of our method

will serve to assess and improve the coverage of the

data until a quality collection is obtained that will

allow the construction of the analysis cube that meets

all the analysis objectives of the project. For example,

if the objective of the analysis is to study the evolution

of opinions about our products of different types of

market participants (e.g. customers, sellers, users

journalists, ...), it will be necessary to retrieve a

representative sample of opinion tweets about each

product in our catalogue, for all the dates of the

analysis period, and for each type of user profile.

In addition, when the analysis cube must be

complete, the availability of metadata becomes an

important issue. For example, some analysis tasks,

such as segmentation of market opinions by gender,

age, location, or profession, require extraction of

metadata from tweets and user descriptions.

However, Twitter users do not always provide these

parameters and some tweets will not be valid because

they lack some key dimensional values. This issue

may mean that the SoBI project is not feasible and

that its analysis objectives should be modified by

redefining or eliminating some attributes or

dimensions of the analysis cube. For example, in most

cases, the analysis of the opinions of consumers must

be executed at country level due to the lack of precise

metadata with the geo-location of the users.

5 CONCLUSIONS

Up to our knowledge, this is the first general approach

to the construction of social media data collections

oriented towards corporate BI tasks. With respect to

previous work, the main contribution of this

methodology is that it considers collection

construction as an iterative exploration process in

which the user analyses the current collection from

the point of view of the analysis objectives and

discovers clues about how to improve it. This data

exploration & profiling task has not received enough

attention in previous methodologies and projects and,

however, in our experiments we have validated that it

helps users to identify the retrieval conditions and

cleaning operations that the construction of each

specific collection requires, as well as, to assess the

feasibility of the analysis objectives of the SoBI

project with respect to the availability of reliable

social media data (Lanza-Cruz et al., 2018) (Berlanga

et al., 2019) (Aramburu et al., 2020).

For future work, an open issue is how to

complement the here presented processing method

with intelligent tools that guide the user in the

selection of the best tweet quality metrics and criteria

for each specific data domain and application. How to

combine them into global quality indicators for a

collection with respect to the analysis objectives of a

SoBI project would be the next step (Berlanga et al.,

2019). New SoBI application scenarios are also

interesting for future research (Aramburu et al.,

2020).

ACKNOWLEDGEMENTS

This research has been funded by the Spanish

Ministry of Industry and Commerce grant number

TIN2017-88805-R and by the pre-doctoral grant of

the Universitat Jaume I with reference

PREDOC/2017/28.

REFERENCES

Abu-Salih, B., Wongthongtham, P., Beheshti,S., Beheshti,

B. (2015). Towards A Methodology for Social Business

Intelligence in the era of Big Social Data incorporating

Trust and Semantic Analysis". In Second International

Conference on Advanced Data and Information

Engineering (DaEng-2015), ed. Bali, Indonesia:

Springer, 2015.

Abu-Salih, B., Chan, K.Y., Al-Kadi, O. et al. Time-aware

domain-based social influence prediction. J Big Data 7,

10 (2020). https://doi.org/10.1186/s40537-020-0283-3.

Agostino, D., Arnaboldi, M., & Azzone, G. (2018). Social

Media Data into Performance Measurement Systems:

Methodologies, Opportunities, and Risks. Azevedo, E.

et al. (Eds.), Handbook of Research on Modernization

and Accountability in Public Sector Management. IGI

Global. 254-275.

Aramburu, M.J.; Berlanga, R.; Lanza, I. (2020) “Social

Media Multidimensional Analysis for Intelligent Health

Surveillance”. Int. J. Environ. Res. Public Health, 17,

2289, https://doi.org/10.3390/ijerph17072289.

Berlanga, R., García-Moya, L., Nebot, V., Aramburu, M. J.,

Sanz, I., & Llidó, D. M. (2015). SLOD-BI: An Open

Data Infrastructure for Enabling Social Business

Intelligence. International Journal of Data Warehousing

and Mining (IJDWM), 11(4), 1-28.

doi:10.4018/ijdwm.2015100101

ICEIS 2021 - 23rd International Conference on Enterprise Information Systems

326

Berlanga, R., Lanza-Cruz I. & Aramburu, M. J. (2019)

"Quality Indicators for Social Business Intelligence,"

2019 Sixth International Conference on Social

Networks Analysis, Management and Security

(SNAMS), Granada, Spain, 2019, pp. 229-236, doi:

10.1109/SNAMS.2019.8931862.

Francia M., Gallinucci E., Golfarelli M., Rizzi S. (2016)

Social Business Intelligence in Action. In: Nurcan S.,

Soffer P., Bajec M., Eder J. (eds) Advanced

Information Systems Engineering. CAiSE 2016.

Lecture Notes in Computer Science, vol 9694.

Springer, Cham.

Gallinucci, Golfarelli & Rizzi, (2015). Advanced topic

modeling for social business intelligence. Information

Systems., 53, pp. 87-106.

García-Moya, L., Anaya-Sánchez, H., Berlanga, R.,

Aramburu, M.J. (2011) Probabilistic Ranking of

Product Features from Customer Reviews. In: Vitrià J.,

Sanches J.M., Hernández M. (eds) Pattern Recognition

and Image Analysis. IbPRIA 2011. Lecture Notes in

Computer Science, vol 6669. Springer, Berlin,

Heidelberg. https://doi.org/10.1007/978-3-642-21257-

4_26.

Goonetilleke, O., Sellis, T., Zhang, X., & Sathe, S. (2014).

Twitter analytics: a big data management perspective.

ACM SIGKDD Explorations Newsletter, 16(1), 11-20.

Gupta, A., Kumaraguru, P., Castillo, C. & Meier, P. (2014).

TweetCred: Real-Time Credibility Assessment of

Content on Twitter. Proceedings of the 6th International

Conference on Social Informatics. 228-243.

10.1007/978-3-319-13734-6_16.

Immonen, A., Pääkkönen, P. & Ovaska, E. (2015).

Evaluating the Quality of Social Media Data in Big

Data Architecture. IEEE Access. 3. 1-1.

10.1109/ACCESS.2015.2490723.

Keegan, B. & Rowley, J. (2017). Evaluation and decision-

making in social media marketing. Management

Decision. 55. 15-31. 10.1108/MD-10-2015-0450.

Lanza-Cruz, I.; Berlanga, R.; Aramburu, M.J. (2018).

Modeling Analytical Streams for Social Business

Intelligence. Informatics. 5. 33. https://doi.org/10.

3390/ informatics5030033

Lee, I. (2018). Social media analytics for enterprises:

Typology, methods, and processes. Business Horizons.

61.2. 199-210, https://doi.org/10.1016/j.bushor.2017.

11.002.

Pääkkönen, P. & Jokitulppo, J. (2017). Quality

management architecture for social media data. J Big

Data 4, 6 (2017). https://doi.org/10.1186/s40537-017-

0066-7.

Pasi, G., Viviani,M. Carton,A. (2019). A Multi-Criteria

Decision Making approach based on the Choquet

integral for assessing the credibility of User-Generated

Content. Information Sciences, vol 503. Pages 574-588.

https://doi.org/10.1016/j.ins.2019.07.037.

Păvăloaia, V.-D., Anastasiei, I.-D., & Fotache, D. (2020).

Social Media and E-mail Marketing Campaigns:

Symmetry versus Convergence. Symmetry, 12(12),

1940. doi:10.3390/sym12121940.

Pereira,V., Fileto, R., Santos,W., Wittwer,M., Reinhold, O.,

Alt, Rainer (2018). A Semantic BI Process for

Detecting and Analyzing Mentions of Interest for a

Domain in Tweets. WebMedia '18: Proceedings of the

24th Brazilian Symposium on Multimedia and the Web.

197–204, https://doi.org/10.1145/3243082.3243100.

Ruhi, U. (2014). Social Media Analytics as a BI Practice:

Current Landscape & Future Prospects. Journal of

Internet Social Networking & Virtual Communities. 1-

12. 10.5171/2014.920553.

Sadiq, S. & Indulska, M. (2017). Open data: Quality over

quantity. International Journal of Information

Management. 37. 150-154. 10.1016/j.ijinfomgt.2017.

01.003.

Sikdar, S., Kang, B., ODonovan, J., Höllerer, T. and Adah,

S. (2013). Understanding Information Credibility on

Twitter, International Conference on Social

Computing, Alexandria, VA, pp. 19-24, doi:

10.1109/SocialCom.2013.9.

Stieglitz, S., Mirbabaie, M., Ross, B., & Neuberger, C.

(2018). Social media analytics - Challenges in topic

discovery, data collection, and data preparation.

International Journal of Information Management. 39.

156-168.

Taleb,I., Serhani, M. A. & Dssouli, R. (2018), Big Data

Quality: A Survey, IEEE International Congress on Big

Data (BigData Congress), San Francisco, CA, pp. 166-

173, doi: 10.1109/BigDataCongress.2018.00029.

Taleb,I. Dssouli, R. & Serhani, M.A. (2015), Big data pre-

processing: A quality framework. 2015 IEEE

international congress on big data. 191-198.

Viviani, Marco & Pasi, Gabriella. (2016). Quantifier

Guided Aggregation for the Veracity Assessment of

Online Reviews. International Journal of Intelligent

Systems. 32. 10.1002/int.21844.

Quality Management in Social Business Intelligence Projects

327