Measurement of Trustworthiness of the Online Reviews Dipankar Das Abstract

2025-05-02 0 0 1.69MB 36 页 10玖币

侵权投诉

Measurement of Trustworthiness of the Online Reviews

Dipankar Das*

Abstract

In electronic commerce (e-commerce)markets, a decision-maker faces a sequential choice problem. Third-

party intervention is essential in making purchase decisions in this choice process. For instance, while purchasing

products/services online, a buyer’s choice or behavior is often affected by the overall reviewers’ ratings, feedback,

etc. Moreover, the reviewer is also a decision-maker. After purchase, the decision-maker would post their reviews

for the product online. Such reviews would affect the purchase decision of another potential buyer, who would

read the reviews before conforming to their ﬁnal purchase. The question that arises is how trustworthy are these

review reports and ratings? The trustworthiness of these review reports and ratings is based on whether the

reviewer is rational or irrational. Indexing the reviewer’s rationality could be a way to quantify a reviewer’s

rationality, but it does not communicate the history of their behavior. In this article, the researcher aims to derive

a rationality pattern function formally and, thereby, the degree of rationality of the decision-maker or the reviewer

in the sequential choice problem in the e-commerce markets. Applying such a rationality pattern function could

make quantifying the rational behavior of an agent participating in the digital markets easier. This, in turn,

is expected to minimize the information asymmetry within the decision-making process and identify the paid

reviewers or manipulative reviews.

keyword: Sequential Choice Problem, Rationality, Pattern, Graph, E-Commerce, Information.

JEL Classiﬁcation Code: D

1 Introduction

The present paper, for the ﬁrst time, gives a measure to understand the irrational online reviews of consumers. The

review comments are about voting against or in favor of the product. Hence, eliminating faulty review comments

is essential in selling the product, setting the prices and advertisements, determining the market-clearing prices,

and the industry competition. Choice models, graph theories, and fuzzy logic are used in the paper to measure the

degree of rationality of the reviewers. Market analysts, marketing managers, and platform economy agents are the

intended audiences.

Third-party intervention in the electronic-commerce(e-commerce) markets plays a vital role in purchasing de-

cisions. By that, the researcher implies reviewer’s ratings, comments, etc., that could affect a buyer’s behav-

ior. The decision to buy a product from an array or list of options often depends upon the number of positive

*Assistant Professor, Goa Institute of Management, India; Email: dipankar@gim.ac.in; dipankar3das@gmail.com

arXiv:2210.00815v2 [econ.TH] 18 Nov 2023

vis-à-vis negative customer comments corresponding to each option. Product ranking through online reviews

based on evidential reasoning theory and stochastic dominance has been studied in [Qin and Zeng, 2022]. Re-

viewer’s attitude towards a speciﬁc topic, or a product, is positive, negative, or neutral has been studied in

[Punetha and Jain, 2023]. The literature proposes that it is essential to quantify the trustworthiness of the re-

viewer’s feedback [Huang and Chen, 2006, Chen et al., 2017, Filieri, 2016, Racherla and Friske, 2012, Utz et al., 2012,

Song et al., 2020]. Ratings also play an essential role, such as an average number of ratings[Khopkar and Nikolaev, 2017,

Saumya et al., 2018, Xu et al., 2012]. Distribution of online ratings plays a crucial informational role in e-commerce

platforms[Etumnu et al., 2020]. Many papers have doubted the reviewer’s ratings and comments and tried quanti-

fying the impacts, including [Koh et al., 2010, Ma et al., 2019, Malik and Hussain, 2018, Pan et al., 2018]. Some

studies have studied linguistics and psychology and combined them with the features of online reviews; such

papers are [Hong et al., 2020, Cui et al., 2020]. Mining online reviews have become an essential tool for identi-

fying consumer behavior and the innovation direction of products. It is difﬁcult for producers and consumers to

analyze and extract relevant opinions from many online reviews effectively. To overcome this, a product ranking

method that combines feature–opinion pairs mining and interval-valued Pythagorean fuzzy sets was proposed in a

study [Fu et al., 2020]. Even though a set of studies exists, these studies could not identify the route of a rational

reviewer or their trustworthiness. No such standard theoretical and quantiﬁable methods are available. Notably,

the reviewer and the decision-maker are both the same person. After a particular purchase, the decision-maker

provides review comments on the digital platform. But, these review comments do not reveal the rationality of

the decision-maker. The condition of rationality interprets the decision-maker’s behavior. The review reports of a

rational decision-maker signiﬁcantly affect the intended decision-maker, who reads the reviews before purchasing.

Indexing the reviewer’s rationality could be a way to quantify it, but indexing does not convey the history of an

agent’s behavior. A consumer could often build up ideas about a product from YouTube review videos (cosmetics,

electronic goods, salon services, etc.). But how does a potential buyer determine the rationality of the YouTuber?

One way to determine the extent of the YouTuber’s rationality would be to review their past uploads. If any of

the YouTuber’s past uploads match the buyer’s purchasing experience, the buyer could also rely on the YouTuber

for future purchases. This paper works with the idea that if it is found that the YouTuber is rational, then the

uploaded videos would get the highest ratings and viewers. This paper gives a standard theory and quantiﬁable

ideas by measuring the new rationality axiom and information indexing about a product and by measuring the

reviewer’s/consumer’s consistency or rationality by using the pattern of past behavior. All reviewers are also

consumers on any given platform, such as Amazon. Therefore, if it is possible to extract the system data of the

records, it would be easier to identify the rationality of a consumer/reviewer.

Rubinstein & Salant, [Rubinstein and Salant, 2006] explain a model of choice from lists where the agent does not

have a comprehensive set of elements before him. Instead, the elements come to their mind sequentially. When

an agent starts selecting from the list or the sequence since, each option carries some information about different

objects. Consider a customer selecting an electronic gadget (say, a cell phone) from the list of cell phones on an

online retail platform (say, Amazon), and the product carries an average rate from reviewers. The concerned agent

reads this information and stops searching for the product for which the information is the maximum (which has

been tested in the paper). There are different information indexing on different digital platforms. Netﬂix uses

a matching index (in percentages), where a particular movie or web series matches with the agent’s preference

expressed in percentages. Other platforms use an average of the reviewers’ ratings on a ﬁve-point scale. The

question lies in ﬁnding a scientiﬁc method to prepare these indices. Will the indices be adjusted according to the

rational judgment of the agent or reviewers? In the e-commerce market, reviewers’ reports and ratings are com-

mon, but it does not reveal the trustworthiness of the reviewers. And the only way to measure trustworthiness is to

measure the degree of rationality of the reviewers. If the pattern of rationality could be attached to the comments,

then a potential buyer would take the decision-making with full information. The information index could be

measured correctly. The present paper formally derives the measure of the pattern of the rationality of an agent in

the e-commerce markets.

Given the existing indices (as mentioned in the above paragraph), the question arises- What are the prerequisites

for an agent to believe the reviewers’ reports and their ratings? Are they rational individually and collectively

so that the ratings can be taken conﬁdently? Information about the consistency of the reviewer is missing. For

example, if a reviewer gives positive (negative) feedback for a particular product, it must imply that the reviewer

consistently prefers (rejecting) that object over other available options. In other words, the reviewer’s preference

is transitive and acyclic at any given time. Whereas, if the reviewer gives negative feedback about a product and

then purchases the product at any given time, they display inconsistent preference towards that object. Would

this be an irrational behavior? Measuring rationality across time is dynamic. It is essential, yet undeﬁned, to de-

rive the rationality pattern dynamically. To begin with, each reviewer has two membership functions: the degree

of consistency and the degree of inconsistency. If these two can be added along with the aggregate preferences

dynamically, then the sequential choice problem would be free from asymmetric and incomplete information.

Moreover, a reviewer gives a review for a particular object. Hence, the rationality of that reviewer towards that

object is essential. The paper derives the pattern of the rationality function on a speciﬁc thing and the overall

rationality in a choice problem in different combinations of objects. Here lies the source of measuring the degree

of rationality.

The design of the paper is as follows. In the ﬁrst segment, the paper derives the choice problem in the e-

commerce market, followed by a discussion on how to read information and index the available information for an

object coming sequentially in the second segment. The third segment narrates a detailed theory of the rationality

pattern function of an agent. The paper concludes by propounding a measure of the degree of rationality to a

particular object and proposing an overall rationality measure.

1.1 Statement of Intended Contribution

Let’s start with a hypothetical example in Table: Review Comments and Ratings of the Given Decision Maker, to

be analyzed in the recent article. The example is based on real-life practices of one agent and two periods model

of choice. For reference see in Figure 1,Figure 2. The real-life review ratings with comments are given in Figure

1& in Figure 2 as examples.

A decision-maker who has taken decisions to select one object from the set Xof four objects X={M, N, V, Z}

Figure 1: Review Rating-Example from amazon.in

in two periods t= 1,2. Now, the article assumes that the two preference patterns given have been extracted

from the model of the choice process, as mentioned in section 3.1. The preference pattern of the given decision

maker at time t= 1 can be written as M→N→V→Z, and the preference pattern at time t= 2 can be

written as Z→V→N→M. Here (→)means preferred to. These two are known and can be extracted

from history. The decision-maker also gives reviews in words with ratings on a ﬁve-star scale for M, N, V, and

Z objects, respectively, in period t= 2 as below in Table: Review Comments and Ratings of the Given Decision

Maker.

Table: Review Comments and Ratings of the Given Decision Maker

Commodity/Object type Review Comments Review Ratings Degree of Trustwor-

thiness

M Bad product ⊛⃝ ⃝ ⃝ ⃝ ?

N Not so good product ⊛ ⊛ ⃝⃝⃝ ?

V Relatively good

product

⊛⊛⊛⃝ ⃝ ?

Z Premium product ⊛⊛⊛⊛⊛ ?

Or, Z Not a Premium prod-

uct

⊛ ⊛ ⊛ ⊛ ⃝?

Hence, the question is to measure the degree of trustworthiness of this hypothetical decision maker (or reviewer).

The last column needs to include the true value of the degree of trustworthiness of each review comment and

the ratings. For example, for object N, the reviewer’s comment is Not so good product with a two-star rating.

But how far these comments and ratings can be trusted? This is not known. Hence, the last column has a

question mark (?). This is why question marks are there for each review’s comments and rating. If this degree of

trustworthiness is attached to it, then the decision-making of the new buyers will be error-free and rational with

complete information. The article derives the method of measuring the degree of trustworthiness of the individual

Figure 2: Review Comments-Example from amazon.in

review comments and ratings.

To understand why this measure of trustworthiness is required, consider an example. A seller knows an object has

demand but posts it at a price higher than the estimated private value. The potential buyer may look for a price

reduction. But if there are review ratings and comments in favor of the object relative to the available substitutes

by previous users, the potential buyer would be tempted to buy the thing at the offered price. This is even at a

price higher than the available alternatives. This implies that review ratings and comments can change private

values further[Das, 2021]. The digital markets, both formal and informal, have an inbuilt system of getting live

reactions from past users of the object being posted for sale. Consumers need complete information about the

prices of goods. Still, their information could be better about the quality variation of objects simply because the

latter statement is more difﬁcult to obtain. The buyer can also buy the thing at a higher price than the private

value. Without any other information, the consumer would not know if he was better off experimenting with low-

or high-priced brands. Consumer behavior is also relevant to determining monopoly power in consumer industries

[Nelson, 1970]. The information problem is to evaluate the utility of each option. Search plays an important

role here. Search to include any way of assessing these options subject to two restrictions: (1) the consumer

must inspect the option, and (2) that inspection must occur before purchasing the brand. Stigler has developed

a theory of search already [Stigler, 1961]. The model is appropriate for the following conditions. Suppose a

consumer has to decide on the number of searches he will undertake before searching. After searching, he can

choose the best from the set of alternatives he has examined. Assume further that he must search by random

sampling and that he knows the form of the probability distribution of his options. After using a brand, its price

and quality can be combined to give us posterior estimates of the utility of its purchase. Digital markets today

have been able to eliminate these shortcomings and have been able to provide posterior estimates of the utility

of its purchase. This is generated in the form of customer reviews, ratings, reactions, comments, etc...Not only

that, this posterior estimate of the utility is not constant but is changing sequentially. Therefore, the expected

文档加载中……请稍候！
如果长时间未打开，您也可以点击刷新试试。

下载文档到电脑，查找使用更方便

10 玖币 0人已下载

立即下载

摘要：

MeasurementofTrustworthinessoftheOnlineReviewsDipankarDas*AbstractInelectroniccommerce(e-commerce)markets,adecision-makerfacesasequentialchoiceproblem.Third-partyinterventionisessentialinmakingpurchasedecisionsinthischoiceprocess.Forinstance,whilepurchasingproducts/servicesonline,abuyer’schoiceorbeh...

展开>> 收起<<

Measurement of Trustworthiness of the Online Reviews Dipankar Das Abstract.pdf

共36页,预览5页

还剩页未读，继续阅读

声明：本站为文档C2C交易模式，即用户上传的文档直接被用户下载，本站只是中间服务平台，本站所有文档下载所得的收益归上传人(含作者)所有。玖贝云文库仅提供信息存储空间，仅对用户上传内容的表现方式做保护处理，对上载内容本身不做任何修改或编辑。若文档所含内容侵犯了您的版权或隐私，请立即通知玖贝云文库，我们立即给予删除！

Measurement of Trustworthiness of the Online Reviews Dipankar Das Abstract

相关推荐

开通VIP享超值会员特权

作者详情

相关内容

热门标签

举报选择: