PeerTrust: supporting reputation-based trust for peer-to-peer electronic communities

Li XiongLing Liu

article2004TKDE1,953 citations

Develops a decentralized, metric-driven reputation framework for peer-to-peer networks that dynamically quantifies node trustworthiness by accounting for feedback credibility, transaction context, and community-specific factors.

Listen

Decentralized peer-to-peer networks and open online marketplaces offer substantial operational flexibility and direct interaction, but their inherent anonymity creates severe vulnerabilities. Without centralized oversight, these environments are frequently exposed to security threats, malicious participants, and fraudulent transactions. Establishing reliable trust is critical, yet traditional centralized rating mechanisms are vulnerable to dishonest feedback, collusion, strategic manipulation, and single points of failure.

The article sets out to design, implement, and validate a decentralized, reputation-based trust framework that accurately quantifies member trustworthiness while resisting malicious feedback and dynamic cheating behaviors.

To accomplish this, the authors introduced an adaptive framework that evaluates trustworthiness through five key parameters: transaction feedback, total transaction volume, the credibility of the rating source, transaction-specific context (such as financial scale), and community-level incentives. The design uses distributed data structures to store reputation information across the network and employs public-key cryptography to ensure data integrity and prevent unauthorized tampering. The authors evaluated the approach using simulated network environments of over one hundred interacting members across thousands of transactions, testing both non-collusive and collusive threat scenarios as well as opportunistic behavioral shifts.

The analysis produced several vital findings. First, weighting feedback by personalized source credibility dramatically outperforms traditional averaging methods, maintaining near-zero computation error and high transaction success rates even when malicious actors represent a significant portion of the network. Second, under coordinated collusive attacks where dishonest members artificially inflate each other's ratings, conventional reputation systems and simple trust metrics fail completely (yielding a 0% transaction success rate), whereas personalized credibility measures successfully neutralize the collusion. Third, incorporating an adaptive time window enables the system to detect and penalize sudden drops in honest behavior far faster than standard fixed-window approaches, making reputation exploitation unprofitable. Finally, utilizing local caching mechanisms allows decentralized trust lookups to scale efficiently with logarithmic network overhead rather than linear complexity.

These findings demonstrate that digital trust frameworks cannot treat all user ratings equally; incorporating source credibility and transaction context is essential to prevent system exploitation. For organizational leaders and platform architects, adopting this multi-factor decentralized trust architecture mitigates operational risks, reduces fraudulent activity, and improves transaction success without requiring costly centralized management.

Organizations operating or designing decentralized networks should implement personalized credibility weighting and adaptive evaluation windows rather than basic cumulative rating scores. They should also integrate lightweight cryptographic signing and localized caching to balance transaction security with operational performance. Future deployments should focus on addressing remaining risks, such as preventing one-time attacks by previously honest actors and managing identity resets.

While the simulation results provide high confidence in the framework's mathematical robustness against dishonest feedback and collusion, practical deployment decisions should recognize certain boundaries. The model assumes participants retain consistent cryptographic identities and does not fully eliminate the threat of compromised nodes or abrupt, one-time exit scams by previously reputable participants.

Georgia Institute of Technology feedback systems online communities p2p environment PeerTrust model reputation-based trust peer-to-peer electronic communities

Xiong et al (2004).pdf

Paper: The Eigentrust algorithm for reputation management in P2P networks, Sepandar D. Kamvar et al. (2003). EigenTrust establishes a distributed P2P reputation baseline whose global trust aggregation PeerTrust adapts with personalized credibility, transaction context, and defenses against collusion.

No sufficiently relevant recommendations were found.

Cover for PeerTrust: supporting reputation-based trust for peer-to-peer electronic communities

Abstract

Peer-to-peer (P2P) online communities are commonly perceived as an environment offering both opportunities and threats. One way to minimize threats in such communities is to use community-based reputations which help to estimate the trustworthiness of peers. This paper presents PeerTrust—a reputation-based trust supporting framework, which includes a coherent adaptive trust model for quantifying and comparing the trustworthiness of peers based on a transaction-based feedback system, and a decentralized implementation of such a model over a structured P2P network. PeerTrust model has two main features. First, we introduce three basic trust parameters and two adaptive factors in computing trustworthiness of peers, namely, feedback a peer receives from other peers, the total number of transactions a peer performs, the credibility of the feedback sources, transaction context factor, and the community context factor. Second, we define a general trust metric to combine these parameters. Other contributions of the paper include strategies used for implementing the trust model in a decentralized P2P environment, evaluation mechanisms to validate the effectiveness and cost of PeerTrust model, and a set of experiments that show the feasibility and benefit of our approach.

Topics

feedback systems online communities p2p environment structured p2p structured p2p networks trust relationship trust evaluation unstructured p2p distributed hash tables

Keywords

PeerTrust model reputation-based trust peer-to-peer electronic communities trust-based peer selection scheme trust metrics trust evaluation reputation management P2P trust

1 INTRODUCTION
2 APPLICATION SCENARIOS AND RESEARCH CHALLENGES
3 THE TRUST MODEL
3.1 Trust Parameters
3.2 General Trust Metric
3.3 The Basic Metric
3.4 Adapting the Trust Metric with Context Factors
4 IMPLEMENTATION STRATEGIES
4.1 Managing Trust Data: System Architecture
4.2 Trust Computation
4.3 Dealing with Dynamic Personality of Peers
4.4 Secure Processing and Transmission of Trust Data
4.5 Trust-Based Peer Selection Scheme
5 EXPERIMENTAL EVALUATION
5.1 Simulation Setup
5.2 Effectiveness against Malicious Behaviors of Peers
5.3 Benefit of the Trust-Based Peer Selection
5.4 Effectiveness against Dynamic Personality of Peers
5.5 Trust Evaluation Cost
6 DISCUSSION
7 RELATED WORK
8 CONCLUSION
ACKNOWLEDGMENTS
REFERENCES

Knowls

Knowl 1 — PeerTrust General Trust Metric
equation
In the PeerTrust framework, the overall trustworthiness $T(u)$ of a peer $u$ within a given recent time window is evaluated using a general trust metric combining five basic parameters:

$T(u) = \alpha \sum_{i=1}^{I(u)} S(u, i) \cdot Cr(p(u, i)) \cdot TF(u, i) + \beta \cdot CF(u)$

where:
- $I(u) \in \mathbb{N}$ denotes the total number of transactions performed by peer $u$ with all other peers within the time window.
- $p(u, i)$ denotes the partner peer participating in peer $u$ 's $i$ -th transaction.
- $S(u, i) \in [0, 1]$ is the normalized degree of satisfaction peer $u$ received from peer $p(u, i)$ for its $i$ -th transaction.
- $Cr(v) \in [0, 1]$ is the credibility factor of the feedback submitted by peer $v$ .
- $TF(u, i) \in [0, \infty)$ is an adaptive transaction context factor reflecting transaction-specific properties such as size or criticality.
- $CF(u) \in [0, \infty)$ is an adaptive community context factor accounting for community-specific incentives or authority measures.
- $\alpha, \beta \in [0, 1]$ are normalized weighting factors satisfying $\alpha + \beta = 1$ that balance collective feedback evaluations against community-level contexts.
Knowl 2 — Trust-Value-Based Credibility Metric
equation
The Trust Value Metric (TVM) models feedback credibility by assuming that trustworthy peers are more likely to submit honest feedback. Omitting transaction and community context factors ( $\alpha = 1, \beta = 0, TF(u, i) = 1$ ), the trustworthiness $T_{\text{TVM}}(u)$ of peer $u$ is computed recursively by weighting each transaction's satisfaction score by the normalized global trust value of the rater:

$T_{\text{TVM}}(u) = \sum_{i=1}^{I(u)} S(u, i) \cdot \frac{T(p(u, i))}{\sum_{j=1}^{I(u)} T(p(u, j))}$

where:
- $I(u)$ is the total number of transactions performed by peer $u$ .
- $p(u, i)$ is the rater peer for peer $u$ 's $i$ -th transaction.
- $S(u, i) \in [0, 1]$ is the normalized satisfaction score received by peer $u$ in transaction $i$ .
- $T(p(u, i))$ is the global trust value of peer $p(u, i)$ .
Because rater trust values $T(p(u, i))$ themselves recursively depend on ratings submitted across the community, $T_{\text{TVM}}$ is evaluated iteratively over all peers until the global trust vector converges.
Knowl 3 — Personalized Similarity Metric for Feedback Credibility
equation
The Personalized Similarity Metric (PSM) computes the trust value $T_{\text{PSM}}(u, w)$ of a target peer $u$ from the personalized perspective of an evaluating peer $w$ . Feedback from a rater $p(u, i)$ is weighted by the historical feedback similarity between $w$ and $p(u, i)$ :

$T_{\text{PSM}}(u, w) = \sum_{i=1}^{I(u)} S(u, i) \cdot \frac{\text{Sim}(p(u, i), w)}{\sum_{j=1}^{I(u)} \text{Sim}(p(u, j), w)}$

where the personalized feedback similarity $\text{Sim}(v, w) \in [0, 1]$ between peer $v$ and peer $w$ is defined over their shared interaction set $IJS(v, w)$ as:

$\text{Sim}(v, w) = 1 - \sqrt{\frac{\sum_{x \in IJS(v, w)} \left(\frac{\sum_{i=1}^{I(x, v)} S(x, i)}{I(x, v)} - \frac{\sum_{i=1}^{I(x, w)} S(x, i)}{I(x, w)}\right)^2}{|IJS(v, w)|}}$

In this formulation:
- $IJS(v, w) = IS(v) \cap IS(w)$ is the set of common peers that have interacted with both peer $v$ and peer $w$ , where $IS(p)$ is the set of all peers that have interacted with peer $p$ .
- $I(x, v)$ is the number of transactions performed between peer $x$ and peer $v$ .
- $S(x, i)$ is the satisfaction rating given to peer $x$ in its $i$ -th transaction.
- $S(u, i)$ is the satisfaction rating given to peer $u$ by rater $p(u, i)$ in peer $u$ 's $i$ -th transaction.
Knowl 4 — Context Factor Adaptations for Transaction Size and Rating Incentives
model/method
The PeerTrust framework adapts the general trust metric to address specific operational and economic risks via transaction and community context factors:
1. Transaction Size Context Factor: To counter subtle malicious strategies where a peer builds a favorable reputation through many minor transactions and defects on large transactions, feedback is scaled by transaction size $D(u, i) > 0$ :
$T(u) = \sum_{i=1}^{I(u)} S(u, p(u, i)) \cdot Cr(p(u, i)) \cdot D(u, i)$

where $S(u, p(u, i))$ is the satisfaction rating for transaction $i$ , $Cr(p(u, i))$ is rater credibility, and $D(u, i)$ is the monetary or critical weight of transaction $i$ .
1. Feedback Incentive Community Context Factor: To alleviate the free-rider problem and encourage active rating, a community reward term is added based on the ratio of feedback provided by peer $u$ ( $F(u)$ ) to peer $u$ 's total transactions ( $I(u)$ ):
$T(u) = \alpha \sum_{i=1}^{I(u)} S(u, i) \cdot Cr(p(u, i)) + \beta \cdot \frac{F(u)}{I(u)}$

where $\alpha + \beta = 1$ . The weighting factors $\alpha$ and $\beta$ control the maximum proportion of reputation achievable through rating participation relative to transaction performance.
Knowl 5 — Adaptive Time-Window Trust Computation Algorithm
algorithm
To detect and mitigate dynamic peer personalities—such as peers that build a high reputation and then begin cheating (reputation milking), or peers oscillating between good and bad behavior—PeerTrust implements an adaptive time-window trust computation.
Input: Target peer $u$ , baseline time window $win$ , smaller adaptive window $wins$ , drop threshold $\epsilon$
Output: Final trust value $T(u)$
Feedback = RetrieveFeedback( $u$ , $win$ )
$T(u)$ = ComputeTrustUsingFeedback(Feedback)
$Ts(u)$ = ComputeTrustUsingFeedback(Subset of Feedback within $wins$ )
if $T(u) - Ts(u) > \epsilon$ then
$T(u) = Ts(u)$
end if
return $T(u)$
By comparing the baseline score $T(u)$ over $win$ against the short-term score $Ts(u)$ over $wins < win$ , any rapid drop in recent service quality triggers an immediate downgrade of the peer's reputation to $Ts(u)$ , while rebuilding reputation requires prolonged good performance across the entire baseline window $win$ .
Knowl 6 — Secure Decentralized Trust Data Management and Verification
algorithm
In PeerTrust, reputation records are stored across a decentralized Distributed Hash Table (DHT) where target peer IDs map to storage keys. To defend against identity spoofing, data tampering in storage or transit, and selective dropping by routing nodes, data exchanges use public key cryptography and replica voting.
Input: Target peer $u$ , replication factor $r$ , polling peer private key $SK(w)$ , polling peer public key $PK(w)$ , time window $win$
Output: Trust value $T(u)$
for $j = 1$ to $r$ do
    response = RetrieveFeedbackSecure( $u$ , $PK(w)$ , $win$ )
    Verify digital signature of response using the replica attached public key
    Feedback = Decrypt response with $SK(w)$
    for $i = 1$ to Length(Feedback) do
        Verify digital signature of Feedback[ $i$ ] using the source rater public key
    end for
     $T_j(u)$ = ComputeTrust( $u$ , Feedback)
end for
$T(u)$ = Median( $T_1(u), T_2(u), ..., T_r(u)$ )
return $T(u)$
Every piece of feedback is signed by the source rater's private key $SK(v)$ , ensuring authenticity and integrity. Replicas encrypt responses with the requester's public key $PK(w)$ so intermediate routing peers cannot inspect or selectively discard ratings. Taking the median across $r$ replicas protects against corrupted storage nodes.
Knowl 7 — Dynamic and Approximate Trust Computation Strategies
model/method
PeerTrust provides two execution strategies for its TVM and PSM trust metrics:
1. Dynamic Trust Computation (DTC):
  - TVM/DTC: Peer $w$ initializes trust values to $T_0(v) = T_{\text{default}}$ and iteratively computes the trust vector across all $N$ peers in the network until convergence ( $\|T^{t+1} - T^t\| < \epsilon$ ). This requires retrieving feedback for all $N$ peers, incurring $O(N)$ lookup hop complexity.
  - PSM/DTC: Peer $w$ retrieves feedback for peer $u$ , then dynamically issues direct lookups to every rater of $u$ to obtain their transaction histories and compute pairwise feedback similarities on demand.
2. Approximate Trust Computation (ATC):
  - Each peer maintains a local trust cache $\text{Cache}_T$ and credibility cache $\text{Cache}_{Cr}$ .
  - TVM/ATC: When computing $T(u)$ , peer $w$ looks up each rater's trust value in $\text{Cache}_T$ . Cache hits provide the credibility score directly; misses substitute $T_{\text{default}}$ , eliminating global iterative computation.
  - PSM/ATC: Peer $w$ checks $\text{Cache}_{Cr}$ for existing rater similarity values, issuing lookups to compute similarity only on cache misses.
  - ATC implementations reduce trust computation lookup complexity to $O(\log N)$ DHT routing hops with $N - 1$ units of local cache storage per peer.
Knowl 8 — Comparative Robustness Against Noncollusive and Collusive Adversaries
empirical result
Simulation experiments across a community of $N = 128$ peers comparing PeerTrust TVM and PSM implementations against a conventional rating average (which omits feedback credibility) show:
- Noncollusive Setting: The conventional approach's computation error (root-mean-square error between computed trust and actual cooperation probability) increases linearly with the malicious peer percentage $k$ . TVM/DTC and TVM/ATC maintain near-zero error when $k < 0.5$ , but error jumps to $1.0$ (complete misclassification of good and bad peers) when malicious peers form the majority ( $k > 0.5$ ). PSM/DTC and PSM/ATC maintain low computation error across the full range of $k \in [0, 0.8]$ .
- Collusive Setting: When malicious peers form a collusive ring to boost one another and badmouth outsiders, both the conventional approach and TVM fail completely even at small malicious fractions ( $k \approx 0.1$ ), resulting in a transaction success rate of $0\%$ . PSM/DTC and PSM/ATC filter out collusive ratings due to low feedback similarity between honest raters and colluding raters, sustaining transaction success rates above $85\%$ .
Knowl 9 — Empirical Evaluation of Adaptive Time-Window Against Reputation Milking and Oscillation
empirical result
In simulations tracking dynamic peer personalities over 500 transactions (baseline window $win = 100$ , adaptive window $wins = 20$ ):
- Reputation Milking: When a peer establishes a high trust score ( $\approx 0.85$ ) and abruptly switches to 100% malicious cheating, the basic fixed-window trust metric allows the peer to exploit a lingering high reputation for over 50 transactions before it decays. The adaptive time-window algorithm detects the divergence between long-term and short-term performance, dropping the trust score below $0.1$ within approximately 20 transactions.
- Reputation Oscillation: When a peer alternates between building reputation and cheating, the basic fixed-window metric yields symmetric costs and gains during building and milking phases. Under the adaptive algorithm, the penalty for cheating is applied rapidly, whereas rebuilding reputation requires sustained honest behavior over the full baseline window $win$ , ensuring that the cost of rebuilding reputation exceeds the gains of milking.
Knowl 10 — Scalability and Lookup Overhead of PeerTrust Computation Strategies
empirical result
Evaluations of network lookup overhead (measured in average network hops per trust computation over a P-Grid DHT) demonstrate:
- Scalability with Network Size ( $N$ ): Across network sizes from $N = 10^1$ to $N = 10^4$ , TVM/ATC and PSM/ATC scale logarithmically ( $O(\log N)$ network hops per evaluation), matching single-peer DHT lookup efficiency. PSM/DTC scales logarithmically but incurs an upward translation corresponding to approximately 100 direct lookups to raters. TVM/DTC scales linearly ( $O(N)$ hops), requiring hundreds to thousands of hops as $N$ increases.
- Cache Bootstrapping Overhead: TVM/ATC exhibits zero bootstrapping overhead because uninitialized cache entries use a default trust value $T_{\text{default}}$ . PSM/ATC requires elevated lookup overhead (up to $\sim 100$ hops) during the first 10–20 computations while filling $\text{Cache}_{Cr}$ , after which its cost drops to match TVM/ATC.
Knowl 11 — Vulnerabilities to Cheap Pseudonyms, One-Time Attacks, and Host Compromise
limitation
The PeerTrust framework identifies three fundamental distributed security limitations that reputation scoring alone cannot prevent:
- Cheap Pseudonyms (Sybil / Reentry Attacks): Malicious peers with poor reputation histories can discard their cryptographic identities and rejoin the network under freshly generated key pairs, resetting their rating history unless external identity-binding costs are imposed.
- One-Time Defection Attacks: A strategic peer that consistently cooperates to accumulate a near-perfect reputation score can exploit that high trust to execute an unavoidable, highly profitable single defection on a critical or high-value transaction before negative feedback can register.
- Host Compromise: If an individual node's operating environment is compromised by malware or worms (such as VBS.Gnutella), local data and private signing keys can be corrupted or exfiltrated, bypassing overlay-level cryptographic protocols.

Coverage note — None was omitted; all contributed models, formulas, decentralized algorithms, empirical results, and analytical limitations are fully covered.

References

1.Gnutella, http://www.gnutella.com, 2000.
2.HyperBee, http://www.hyperbee.com, 2001.
3.PeerCQ, http://disl.cc.gatech.edu/PeerCQ, 2004.
4.A. Abdul-Rahman and S. Hailes, “Supporting Trust in Virtual Communities,” Proc. 33rd Ann. Hawaii Int’l Conf. System Sciences (HICSS-33), 2000.
5.K. Aberer, “P-Grid: A Self-Organizing Access Structure for P2P Information Systems,” Proc. Ninth Int’l Conf. Cooperative Information Systems, 2001.
6.K. Aberer and Z. Despotovic, “Managing Trust in a Peer-to-Peer Information System,” Proc. ACM Conf. Information and Knowledge Management (CIKM), 2001.
7.Y. Atif, “Building Trust in E-Commerce,” IEEE Internet Computing, vol. 6, no. 1, 2002.
8.M. Chen and J.P. Singh, “Computing and Using Reputations for Internet Ratings,” Proc. Third ACM Conf. Electronic Commerce, 2001.
9.F. Cornelli, E. Damiani, S.D.C. di Vimercati, S. Paraboschi, and P. Samarati, “Choosing Reputable Servents in a P2P Network,” Proc. 11th Int’l World Wide Web Conf., 2002.
10.C. Dellarocas, “Immunizing Online Reputation Reporting Systems against Unfair Ratings and Discriminatory Behavior,” Proc. Second ACM Conf. Electronic Commerce, 2000.
11.C. Dellarocas, “Analyzing the Economic Efficiency of Ebay-Like Online Reputation Reporting Mechanisms,” Proc. Third ACM Conf. Electronic Commerce, 2001.
12.C. Dellarocas, “The Digitization of Word-of-Mouth: Promise and Challenges of Online Reputation Mechanism,” Management Science, vol. 49, no. 10, 2003.
13.E. Friedman and P. Resnick, “The Social Cost of Cheap Pseudonyms,” J. Economics and Management Strategy, vol. 10, no. 1, 2001.
14.B. Holmstrom, “Managerial Incentive Problems: A Dynamic Perspective,” Rev. Economic Studies, vol. 66, no. 1, 1999.
15.B.A. Huberman and F. Wu, The Dynamics of Reputation. 2002.
16.A. Josang and R. Ismail, “The Beta Reputation System,” Proc. 15th Bled Conf. Electronic Commerce, 2002.
17.A. Josang and N. Tran, “Simulating the Effect of Reputation Systems on E-Markets,” Proc. First Int’l Conf. Trust Management, 2003.
18.S. Kamvar, M. Scholsser, and H. Garcia-Molina, “The EigenTrust Algorithm for Reputation Management in P2P Networks,” Proc. 12th Int’l World Wide Web Conf., 2003.
19.S. Ketchpel and H. Garcia-Molina, “Making Trust Explicit in Distributed Commerce Transactions,” Proc. 16th Int’l Conf. Distributed Computing Systems, 1996.
20.D. Kreps and R. Wilson, “Reputation and Imperfect Information,” J. Economic Theory, vol. 27, 1982.
21.R.A. Malaga, “Web-Based Reputation Management Systems: Problems and Suggested Solutions,” Electronic Commerce Research, vol. 1, no. 4, 2001.
22.D.W. Manchala, “E-Commerce Trust Metrics and Models,” IEEE Internet Computing, vol. 4, no. 2, 2000.
23.D.H. McKnight and N.L. Chervany, “The Meanings of Trust,” Technical Report WP9604, Univ. of Minnesota Management Information Systems Research Center, 1996.
24.N.H. Miller, P. Resnick, and R.J. Zeckhauser, “Eliciting Honest Feedback in Electronic Markets,” KSG Working Paper Series RWP02-039, 2002.
25.L. Mui, M. Mohtashemi, and A. Halberstadt, “Notions of Reputation in Multi-Agent Systems: A Review,” Proc. First Int’l Joint Conf. Autonomous Agents and Multiagent Systems, 2002.
26.J.M. Pujol, R. Sanguesa, and J. Delgado, “Extracting Reputation in Multi-Agent Systems by Means of Social Network Topology,” Proc. First Int’l Joint Conf. Autonomous Agents and Multiagent Systems, 2002.
27.S. Ratnasamy, P. Francis, M. Handley, R. Karp, and S. Shenker, “A Scalable Content Addressable Network,” Proc. ACM SIGCOMM, 2001.
28.P. Resnick, R. Zeckhauser, E. Friedman, and K. Kuwabara, “Reputation Systems,” Comm. ACM, vol. 43, no. 12, 2000.
29.A. Rowstron and P. Druschel, “Pastry: Scalable, Decentralized Object Location and Routing for Large-Scale Peer-to-Peer Systems,” Middleware, 2001.
30.J. Sabater and C. Sierra, “Reputation and Social Network Analysis in Multi-Agent Systems,” Proc. First Int’l Joint Conf. Autonomous Agents and Multiagent Systems, 2002.
31.S. Sen and N. Sajja, “Robustness of Reputation-Based Trust: Boolean Case,” Proc. First Int’l Joint Conf. Autonomous Agents and Multiagent Systems, 2002.
32.I. Stoica, R. Morris, D. Karger, F. Kaashoek, and H. Balakrishnan, “Chord: A Scalable Peer-to-Peer Lookup Service for Internet Applications,” Proc. ACM SIGCOMM, 2001.
33.L. Xiong and L. Liu, “A Reputation-Based Trust Model for Peer-to-Peer Ecommerce Communities,” Proc. IEEE Conf. E-Commerce (CEC ’03), 2003.
34.J.E. Youll, “Peer to Peer Transactions in Agent-Mediated Electronic Commerce,” Master’s thesis, Massachusetts Inst. of Technology, 2001.
35.B. Yu and M.P. Singh, “A Social Mechanism of Reputation Management in Electronic Communities,” Proc. Seventh Int’l Conf. Cooperative Information Agents, 2000.
36.G. Zacharia and P. Maes, “Trust Management through Reputation Mechanisms,” Applied Artificial Intelligence, vol. 14, no. 8, 2000.

Citation

MLA

Li Xiong, and Ling Liu. “PeerTrust: Supporting Reputation-Based Trust for Peer-to-Peer Electronic Communities”. IEEE Transactions on Knowledge and Data Engineering, vol. 16, no. 07, 2004, pp. 843–57, https://doi.org/10.1109/TKDE.2004.1318566.

APA

Li Xiong, & Ling Liu. (2004). PeerTrust: Supporting Reputation-Based Trust for Peer-to-Peer Electronic Communities. IEEE Transactions on Knowledge and Data Engineering, 16(07), 843–857. https://doi.org/10.1109/TKDE.2004.1318566

Chicago

Li Xiong, and Ling Liu. 2004. “PeerTrust: Supporting Reputation-Based Trust for Peer-to-Peer Electronic Communities”. IEEE Transactions on Knowledge and Data Engineering 16 (07): 843–57. https://doi.org/10.1109/TKDE.2004.1318566.

Harvard

Li Xiong and Ling Liu (2004) “PeerTrust: Supporting Reputation-Based Trust for Peer-to-Peer Electronic Communities”, IEEE Transactions on Knowledge and Data Engineering, 16(07), pp. 843–857. Available at: https://doi.org/10.1109/TKDE.2004.1318566.

Vancouver

1. Li Xiong, Ling Liu (2004) PeerTrust: Supporting Reputation-Based Trust for Peer-to-Peer Electronic Communities. IEEE Transactions on Knowledge and Data Engineering 16:843–857

BibTeX

@article{Li_Xiong_2004, title={PeerTrust: Supporting Reputation-Based Trust for Peer-to-Peer Electronic Communities}, volume={16}, ISSN={1041-4347}, url={http://dx.doi.org/10.1109/TKDE.2004.1318566}, DOI={10.1109/tkde.2004.1318566}, number={07}, journal={IEEE Transactions on Knowledge and Data Engineering}, publisher={Institute of Electrical and Electronics Engineers (IEEE)}, author={Li Xiong and Ling Liu}, year={2004}, month=July, pages={843–857} }

Metadata:Crossref

Access the Paper

This paper is available from its original source. Click below to access the PDF.

Open PDF

PeerTrust: supporting reputation-based trust for peer-to-peer electronic communities

Abstract

Topics

Keywords

Table of Contents

Knowls

Knowl 1 — PeerTrust General Trust Metric

Knowl 2 — Trust-Value-Based Credibility Metric

Knowl 3 — Personalized Similarity Metric for Feedback Credibility

Knowl 4 — Context Factor Adaptations for Transaction Size and Rating Incentives

Knowl 5 — Adaptive Time-Window Trust Computation Algorithm

Knowl 6 — Secure Decentralized Trust Data Management and Verification

Knowl 7 — Dynamic and Approximate Trust Computation Strategies

Knowl 8 — Comparative Robustness Against Noncollusive and Collusive Adversaries

Knowl 9 — Empirical Evaluation of Adaptive Time-Window Against Reputation Milking and Oscillation

Knowl 10 — Scalability and Lookup Overhead of PeerTrust Computation Strategies

Knowl 11 — Vulnerabilities to Cheap Pseudonyms, One-Time Attacks, and Host Compromise

References

Citation

BibTeX

Access the Paper