This chapter has presented a literature review and essential background for the research work described in this thesis focusing on: sentiment mining approaches, text visualisa- tion and social network analysis. The review considered both the classification-based approach to sentiment analysis as well as the lexicon-based approaches (generic and domain specific). This chapter also reviewed some related work on sentiment analy- sis in the political domain and has described some related work on visualising debate structures and analysing such structures. The next chapter introduces the datasets used for evaluation purposes with respect to the work presented in the remainder of this thesis.
Chapter 3
The UK House of Commons
Political Debates Corpus
“There is no such thing as public opinion. There is only published opinion.”
Winston Churchill
This chapter presents an overview of the UK House of Commons parliamentary debate datasets which were used to act as a focus for the work described in this thesis
in the context of the evaluation of the techniques proposed. More specifically the
datasets were used for:
• Evaluation purposes with respect to attitude classification in the context of politi-
cal sentiment mining using machine learning classification as described in Chapter
4.
• Evaluation purposes with respect to attitude prediction in the context of political
sentiment mining using generic sentiment lexicons as described in Chapter 5.
• Evaluation purposes with respect to attitude prediction in the context of polit-
ical sentiment mining using the proposed domain specific sentiment lexicons as
described in Chapter6.
• The comparison and contrast purposes with respect to the three approaches to po-
litical sentiment mining considered in Chapters4,5and 6by considering debater
• The extraction of debate graphs embedded within debate transcriptions, using the proposed Debate Graph Extraction (DGE) framework, so as to support the graphical visualisation of the high-level structure of political debates as described
in Chapter8.
• Evaluation purposes with respect to the determination of sentiment scores and
polarities within a pre-labelled corpus, using the proposed ∆TF-IDF0 weighting
scheme, as described in Chapter 6.
• The generation of the two proposed domain specific (political) sentiment lexicons,
PoLex and PoliSentiWordNet, generated by applying the two techniques described
in Chapter6 to political benchmark data.
This chapter commences with an overview of the UK Parliamentry system in Section
3.1 so that the reader can more precisely understand the problem domain. Section
3.2 then provides a brief description of the political party system in operation in the
UK, while Section 3.3 describes the nature and characteristics of UK parliamentary
debates. Section3.4 reviews the UK House of Commons datasets used with respect to
the research described in this thesis. Collectively these datasets were referred to as the UK House of Commons Debate (UKHCD) datasets. In total four UKHCD datasets were used: UKHCD-1, UKHCD-2, UKHCD-3 and UKHCD-4. Finally, this chapter is
completed with a brief summary in Section 3.5.
3.1
The UK Parliamentry System
The UK Parliament1 consists of two Houses: (i) the House of Commons (“The Com-
mons”) and (ii) the House of Lords (“The Lords”). Both houses have similar respon- sibilities which comprise: (i) making laws (legislation), (ii) checking the work of Gov- ernment (scrutiny) and (iii) debating running important political issues. The House of Lords is independent from the elected House of Commons and (generally) any decision made by one House needs to be approved by the other House in a two-chamber system. More precisely:
The Commons: The House of Commons has the most authority and comprises an
elected membership. The UK public elects (currently) some 650 Members of Par- liament (MPs) to represent them in the House of Commons. Almost all MPs are members of political parties and the party with the largest number of MPs in the Commons forms the Government (although at time of writing no party had
1This chapter contains information taken from the UK Parliament official website available on-line at (http://www.parliament.uk/).
sufficient votes to gain overall control, hence the Government comprised a coali- tion). MPs debate running political issues, propose new laws and make decisions on financial Bills and proposed new taxes. Debates in the House of Commons are chaired by what is known as the “Speaker” or one of his/her deputies. These debates are recorded and publicly available. It is these House of Commons de- bates that are of significance with respect to the work described in this thesis and
hence they are discussed in more detail later in this Chapter in Section3.4.
The Lords: The House of Lords is the second chamber of the UK Parliament and has
less authority than The Commons. Members of the House of Lords are appointed by the Queen on the advice of the Prime Minister or recommended by the House of Lords Appointments Commission. The main roles of the House of Lords are to consider proposed legislation and to provide for a check on Government activities. The Lords can propose amendments to legislation, but ultimately cannot block legislation proposed by the House of Commons. The House of Lords also conducts its business through a process of debate and these are also recorded. However these debates are not considered in the context of this thesis, which concentrates on Commons debates although there is no reason why the proposed techniques cannot be equally applied to House of Lords debates.
Thus, so as to provide a focus for the research work described in this thesis UK House of Commons debates were used. To explain how UK House of Commons parliamentary debates are conducted the role of the House of Commons should first be considered in more detail:
1. Monitoring the work of the Government (scrutiny): Parliament monitors the work of the Government by: (i) directing questions to Government ministers, (ii) debating and (iii) constituting investigative committees. In more detail:
• Questions: Minsters may attend Parliament to answer questions orally, or
in writing. The Prime Minister answers questions every Wednesday.
• Debates: Debates in the Commons are conducted so as to make or amend
laws.
• Committees: Committees consist of smaller groups of MPs. Their role is to
look at specific policy issues and/or legislation in detail. Different commit- tees have different roles ranging from offering advice to producing reports suggesting modifications to legislation.
2. Debating and passing of laws (legislation): Parliament is responsible for debating and voting on new laws or modifications to existing laws that are proposed by
the Government or in some cases an MP, a member of the House of Lords or a member of the public. Making Laws is conducted as follows:
• Introducing Bills: A Bill is a proposal to introduce a new law or modifications
to an existing law, which is debated by Parliament.
• Approval: To become Law the text of a Bill must be agreed by both Houses2.
• Royal Assent: The reigning monarch has to approve all new Laws. The
Royal Assent is a formality and is not withheld. Once a Bill is approved by Royal Assent it becomes a Law and the Government is responsible for implementing that Law.
3. Enabling the Government to raise taxes.