In this paper, we describe ongoing work on online banking customization with a particular focus on interaction. The scope of the study is confined to the Australian banking context where the lack of customization is apparent. This paper puts forward the notion of using tags to facilitate personalized interactions in online banking. We argue that tags can afford simple and intuitive interactions unique to every individual in both online and mobile environments. Firstly, through a review of related literature, we frame our work in the customization domain. Secondly, we define our main idea and identify a range of taggable resources in online banking. Thirdly, we describe our preliminary prototype implementation with respect to interaction customization types. Lastly, we conclude with a discussion on future work.
The workshop theme is centered on data-centric interaction on the Web. The everincreasing amount of data available online has lead to a growing interest in exploring its potential use to enhance Web-based interaction. This paper discusses the use of user-contributed data in particular user tags in the online banking space with the aim of delivering a personalized interaction.
This paper is specifically aimed at customization in the online banking context
because it is an imperative dimension of user satisfaction, particularly among the
younger generation in Australia [
The rest of this paper is structured as follows. In section 2, the related work on customization is presented and background on the proposed technology is offered. In section 3, the proposed approach is outlined along with results of a background study to identify user taggable resources in online banking. In section 4, the preliminary prototype implementation is detailed with reference to interaction customization types. In section 5, the paper is concluded with discussion on future work.
2 2.1
Based on the literature, a customization framework has been derived, which offers an
overview of the customization domain, and position of the work described in this
paper (highlighted). For the purpose of literature review, a broad understanding of
customization was adopted defined as “the ability for a website to be shaped in a way
that better fulfils the wants of individual users” [
Customization Category Interaction
Process
User Interface
Content &
Features Layout
Remembering
Information Look & Feel
Comprehension
Link Navigation
Associative
Source
Action Explicit Implicit
User System application of customization, while system initiated and executed customization is commonly described as dynamic application of customization. However, in some instances customization may be initiated by the user and carried out by the system or vice-versa. Arguably, these entities are inseparable and equally important hence their inclusion in some stage of the customization process is advantageous for an inclusive result.
The three types of customization outlined by Fung [
Tags, also known as user-defined metadata are a popular Web 2.0 technology,
enabling users to assign keywords to Web resources (e.g., photo, video, people, etc)
primarily for the purpose of personal information management (PIM). Tags are
largely personal and contextual [
In the financial services space, tags are widely used to assist personal financial management via third party tools such as Mint2 and Yodlee3, where a user can assign tags to annotate transactional data for purposes such as budgeting, expense tracking, etc. However, these tools, allow tags to be assigned to financial transactions at a high level as category or description, but not at a lower level for details such as bank account, biller, description, etc. There may be compelling advantages in doing so in the online banking environment, opening doors to tag-based interactions alongside personal financial management. Although the ability to tag financial data has existed for a while now primarily through third party tools, this trend is likely to change with banks considering the inclusion of personal financial management features. This view is particularly evident among Australian banks who have began to do so as part of their offerings with Australia and New Zealand Bank (ANZ) pioneering the initiative through its ANZ-MoneyManager4 service. The inclusion of tags as part of online banking adds to the relevance and practicality of the proposed approach to customize online banking interaction.
3 3.1
The main idea of our approach is to use user-defined tags and tag-based visualization (tag cloud) to facilitate customization in online banking. Tags and tag clouds can offer a more intuitive and interactive online banking website. Since tags represent resources in a personal manner, tags can be used to facilitate comprehension of user intentions. Possible actions may be inferred from tags based on the resources they represent and the relationships between those resources. Also, relevant tags can be recommended to users and based on semantic relations of tags across the network, related services can be aggregated and recommended, which banking users may find useful. 3.2
In order to define the range of taggable resources in online banking, a piece of information not readily available from the literature, a background case study was conducted. The study involved manual examination of personal banking websites of two leading banks in Australia: Commonwealth Bank5 and Suncorp Bank6. Personal banking was preferred over other types of banking because it appeals to a wider base of users. Both online and mobile banking environments were considered in this study.
To identify taggable resources we observed both aforementioned online banking websites, focusing on services offered and the different information required for the services including fund transfer, bill payment, product application and internal messaging. The results of the study are presented in Table 1. A total of seven resources were identified, grouped into five categories: account, description, biller, application and message.
owned accounts such as every day, savings, cheque, credit card, etc Payee account which are either linked accounts such as personal accounts or other third party accounts that are internal, external or overseas Personal description of a transaction as self-reference. Transaction types include offline such as EFTPOS, direct debit, etc; and online such as bill pay, fund transfer, shopping, etc
R1 R2 R3 Account
R4 R5 R6 R7 Biller
All types of financial products such as account, credit card, loans, etc Personal communication between customer and bank
The findings of the study show that the resources identified are typically available in both environments which suggest that tag-based interaction is applicable to different contexts. However, it is noteworthy that mobile banking to a large extent only supports the use of pre-existing resources rather than new resources. For example, a fund transfer can only be carried out to a saved payee but not to a new payee in which case the payee information has to be added through the online environment. This is possibly due to the inconvenience of entering information through a mobile device or to security and privacy concerns related to the use of mobile banking. 4
An early prototype implementation with tag integration for a few key resources namely account, description and biller has been developed. The prototype is softwarebased, intended to demonstrate the mechanics of each customization particularly in the desktop environment for two main activities: bill payment and fund transfer. Although the prototype is still in its early days, it extends support to the proposed approach especially in terms of feasibility and practicality.
The following sub-sections elaborate the different types of interaction customization and the proposed use of tags to address them. The examples largely depict scenarios of day to day online banking activities. 4.1
This customization type is defined as the ability to provide customization through
simple remembering of a user’s information based on the recurrence rate of a
particular action on a website [
Remembering-type customization can be fulfilled through tags assigned to resources that are presented as tag clouds. This provides a visual retrieval interface that can simplify and ease the execution of past or recurring transactions. Simply by clicking on a tag, related information about a transaction that the tag is associated with can be retrieved and displayed. If a selected tag is associated with two or more tags then the tag cloud can be filtered to show tags which are co-occurring with the selected tag. This removes the need to navigate to a different page or perform a manual search query. This also means for carrying out a past or recurring transaction, users will only need to update necessary information such as amount (if different) and possibly retain other details such as bank accounts and description.
Based on user’s tagged resources namely transaction description, remembrancebased customization is introduced. The following example assumes a user pays a monthly mobile bill and tagged the transaction as “mobile” in the first month, and the following month the user returns to carry out the same activity.
Scenario 1: Mobile bill payment. User clicks on “mobile” tag (1) from the tag cloud. As a result, the bill payment form is automatically completed. *The selected tag (“mobile”) from the tag cloud is highlighted (underline) and the transaction details are loaded and the account and biller tags (“Everyday” & “Vodafone”) are automatically selected (tick icon). 4.2
This customization type is defined as the ability to recognize user’s behaviors and
provide assistance towards fulfilling the user’s needs [
Comprehension-type customization can be fulfilled by inferring banking actions
like fund transfer based only on tags selected by the user. Such inference is possible
particularly for tags with certain types of relations such as two bank accounts. These
relations when combined with simple pre-defined rules can aid in populating relevant
actions. A sample pre-defined rule is the ability to transfer funds from savings
account to credit card account but not the other way around due to nature of the
accounts. However, it would be possible to transfer two-ways between a savings
account and a current account. The default choice for the suggested actions can be
made based on past user actions to closely reflect user's needs. However, as a key
HCI design rule, it is important not to automatically carry out an action to ensure the
locus of control remains with the user [
Based on user’s tagged resources namely accounts and billers, the comprehensionbased customization is detailed. This customization is realized through user selection of tags, where a set of actions are inferred by analyzing the relations between tags selected. This allows users to carry out their banking activities with minimal effort. This is achieved by examining the tripartite structure of tags comprising of user, tag and resource and subsequently, applying pre-defined rules to the underlying resources. Rules are defined for a particular resource owned by a user such as bank account based on account type. For example, a savings account would have three rules: transfer_from, transfer_to and view, which denotes that the account can be used to send or receive money, and be viewed. Meanwhile a credit card account (e.g., “Visa”) would only have two rules: transfer_to and view, which denotes that the account can only receive money and be viewed. Similarly, a payee account (e.g., “Dad”) would also have the same set of rules as the credit card account. The account with ‘transfer_from’ action will always act as the sender/primary account while the account with ‘transfer_to’ action will act as the recipient/secondary account. In the event tags of two accounts with ‘transfer_from’ action are selected, it would be possible to have identical actions of transferring and receiving funds between both accounts. In order to reduce the overall complexity involving multiple accounts, a limit for the number of selectable tags particularly for personal accounts is desirable (refer to Table 1 for account types).
The following examples show the ability of conducting an internal and external fund transfers just by selecting tags. Although the examples only illustrate fund transfers to a single account, it is possible to carry out fund transfers to multiple accounts at once.
Scenario 2: Internal fund transfer from Savings to Everyday account. User clicks on
“Savings” (1) and “Everyday” (2) tags, a set of possible actions for these accounts
are populated. The suggested actions are 1) ‘Transfer from Savings to Everyday’, 2)
‘Transfer from Everyday to Savings’ and 3) ‘View transaction history of Savings and
Everyday’.
Scenario 3: External fund transfer from Everyday to John’s account. User clicks on
“Everyday” (1) and “John” (2) tags, a set of possible actions for these accounts are
populated. The suggested actions are 1) ‘Transfer from Everyday to John’ and 2)
‘View transaction history of Everyday and John’
This customization type is defined as the ability to provide customization through
association of user’s behavior with other individuals who share similar interests or
needs[
Associative-type customization can be fulfilled through tag recommendation to
users primarily as suggestions in the form of dropdown box or visually through the
use of tag clouds to display related tags. The semantics of tags can be used to find
closer association between tags across the network and to select/rank the most
relevant sets of tags based on similarity score [
Based on user’s tagged resources namely biller and description, association-based
customization is explicated. The associations are divided into tag suggestions and
tagbased service aggregation. Both these associations are based on two sets of tags:
personal and public (cross-network). For tag suggestions, personal tags take
precedence over public tags, and the most relevant public tags can be suggested based
on number of associations to a resource. To improve the relevance of tags suggested
to an individual, analysis of semantic relatedness is useful [
The following examples illustrate this customization by assuming a user carries out two select bill payments to two billers: Vodafone (mobile) and OzForex (foreign currency exchange).
Scenario 4: Tag recommendation for multiple bill payment (mobile and money transfer). User clicks on “Vodafone” (1) and then “OzForex” (2) biller tags, and clicks to enter a description tag (3). As a result, a set of related tags are recommended that are used in the context of the selected billers. Scenario 5: Tag-based services recommendation. User clicks on “forex” (1) tag, related services are populated with aggregated information on service usage. This paper firstly provides a conceptual understanding of interaction-focused customization in the larger customization domain. Secondly, it defines the range of user taggable resources in both online and mobile banking environments. Thirdly, a practical view of the proposed tag-based interaction is offered via a preliminary prototype implementation. The main point of interest of the research lies in the potential use of tags to facilitate personalized banking interactions. The preliminary prototype highlights this specifically in the desktop environment. The ability to employ a light-weight user-driven technology such as tags to facilitate various interaction customization types is advantageous and useful, and may be achieved with minimal effort for implementation and adoption. However, a drawback to using tags is the lack of detailed information about resources which they represent that may be needed to ensure correct use of online banking. Even though the ability to recall information as a form of self-communication is significant with tags [11], users may still want to view detailed information about a resource. One possible solution is to provide detailed information about tagged resources on demand in the form of dynamic tooltip as a simple hover effect on desktop and tap and hold action on mobile devices. The use of dynamic tooltips also allows the most up-to-date information such as balance amount to be shown on the fly, which is important for carrying out a transaction.
From here, we envisage further exploration of tags and tag clouds in the mobile
environment, particularly addressing the spatial constraint present. This can highlight
other form of interaction techniques which may help to better achieve the different
customization types. Also, we will focus on evaluating the prototype with the aim of
assessing its utility and usability in both online and mobile environments as they are
key factors of user performance, satisfaction and acceptance [12]. This will allow us
to explore ways in which the prototype can be improved. The proposed method of
evaluation is experimentation using a pretest-posttest control group design. This
design would provide strong internal validity, suitable to measure cause-effect
relationships [13]. Each customization type will be evaluated separately in a similar
fashion to Fung’s study [
This research work is sponsored by the Smart Services Cooperative Research Centre (CRC) of Australia (http://www.smartservicescrc.com.au) as part of the financial services project. We would like to thank our fellow colleague Ms Claudia Murillo for her feedback and also the reviewers of this paper for all their comments and suggestions. 7