A Mathematical Trust Algebra for International Nation Relations Computation and Evaluation
Mohd Anuar Mat Isa, Ramlan Mahmod, Nur Izura Udzir, Jamalul-lail Ab Manan, Ali Dehghan Tanha
A Mathematical Trust Algebra for International Nation Relations Computation and Evaluation
Mohd Anuar Mat Isa (*corresponding author ) Ramlan Mahmod Nur Izura Udzir
Faculty of Computer Science and Information Technology, Universiti Putra Malaysia, Malaysia
Jamalul-lail Ab Manan , Section 7, Shah Alam, Malaysia.
Ali Dehghan Tanha , University of Salford, The Crescent, Salford, United Kingdom.
Abstract
This manuscript presents a trust computation for international relations and its calculus, which related to Bayesian inference, Dempster-Shafer theory and subjective logic. We proposed a method that allows a trust computation which is previously subjective and incomputable. An example of case study for the trust computation is the United States of America–Great Britain relations. The method supports decision makers in a government such as foreign ministry, defense ministry, presidential or prime minister office. The Department of Defense (DoD) may use our method to determine a nation that can be known as a friendly, neutral or hostile nation.
Keywords
Trust Algebra, International Relations, Trust Computation, Foreign Policy, Politics, Dempster-Shafer, Subjective Logic, Common Criteria, Defense, National Security, Terrorism, Counter Terrorism, Trust Perception, Foreign Ministry, Intelligent
Introduction
This publication describes an extension of our previous works related to trust issues in international nation relations. In our previous works (Mohd Anuar Mat Isa et al., 2012a, 2012b), we have mentioned the need for a “ trust model” in Common Criteria (CC). In this work, we model the international relations between nations using a calculus model, which we call a trust algebra. The proposed method will allow a trust computation, which is previously subjective and incomputable.
Related Work
There are many trust definitions that were expressed in natural languages. To enable trust computation in computing systems, the trust definitions need to be conveyed in a measurable or quantifiable notations such as statistical representations. The first attempt to represent trust in the statistical notations was by (Dempster, 1967). Dempster showed a probability measurement that is to define an upper and lower probabilities for a multivalued mapping. The probability measurement is a generalization of calculus in Bayesian theory. His statistical scheme is adopted by (Shafer, 1979, 1976) and it provides elegant method to compute trust. Many researchers later (in 1980-1995) addressed both works as the foundation for a concrete trust computation, which they began to call as Dempster-Shafer theory in the early 1980s (Shafer, This manuscript is a draft version. The final version will be published in a reputable journal. One may contact the main author ([email protected]) for further clarification or discussion for an implementation of the trust algebra.
Trust Algebra: Definition and Notation
Definition 1 . A nation state is a sovereign nation and recognized by the United Nation (UN). Referring to the UN’s Charter (“Charter of the United Nations,” 1945): Chapter I, Articles 1: “
To maintain international peace and security …” and “ to develop friendly relations among nations based on respect for the principle of equal rights …”. Articles 2: “… principle of the sovereign equality of all its Members .”. Chapter II, Articles 4: “
Membership in the United Nations is open to all other peace-loving states which accept the obligations contained in the present Charter …”. Referring to the UN’s Charter, we define a nation term in this work as the nation state or any UN member states.
Definition 2.
Trust relation is a relationship between Nation A and Nation B. The trust relation can be either friendly (ally), neutral, or hostile (enemy). The trust relation ℛ 𝐴,𝐵 denotes a trust perception of Nation A toward Nation B. Assume that: 𝑓 ≝ 𝑓𝑟𝑖𝑒𝑛𝑑𝑙𝑦, 𝑛 ≝ 𝑛𝑒𝑢𝑡𝑟𝑎𝑙, ℎ ≝ ℎ𝑜𝑠𝑡𝑖𝑙𝑒 𝑓, 𝑛, ℎ ∈ 𝑅𝐸𝐿𝐴𝑇𝐼𝑂𝑁
𝐴, 𝐵 ∈ 𝑁𝐴𝑇𝐼𝑂𝑁 ℛ 𝐴,𝐵 = (𝑓 ∩ 𝑛 ∩ ℎ) = ∅ ℛ 𝐴,𝐵 = (𝑓 ∩ 𝑛) ∪ (𝑓 ∩ ℎ) ∪ (𝑛 ∩ ℎ) = ∅ ; Remark 2.1
Trust relation for ℛ 𝐴,𝐴 is reflexive with always friendly.
Remark 2.2
Trust relation for ℛ 𝐴,𝐵 ≠ ℛ
𝐵,𝐴 is not always symmetric.
Remark 2.3
Trust relation for ℛ 𝐴,𝐵 and ℛ 𝐵,𝐶 does not always imply that ℛ 𝐴,𝐶 is transitive for relations between Nation A, Nation B and Nation C.
Remark 2.4
Trust relation for ℛ 𝐴,𝐵 and ℛ 𝐵,𝐴 are commutative for binary operation (ℛ 𝐴,𝐵 , ℛ
𝐵,𝐴 ) = (ℛ
𝐵,𝐴 , ℛ
𝐴,𝐵 ) for addition and multiplication operations. Definition 3 . Trust relation for Nation A and Nation B is undefined for 𝓡 𝑨,𝑩 = (𝒇 ∪ 𝒏 ∪ 𝒉) = ∅ . Remark 3.1
Trust relation for ℛ 𝐴,𝐵 is undefined when a relation between Nation A and Nation B is neither friendly, neutral, nor hostile. The state of the relation is unknown.
Remark 3.2
If a definition of a nation is reduced to Definition 1, the trust relation always exists because of diplomatic relations and recognitions.
Definition 4.
Weightage is used for a linear normalization of trust perceptions between Nation A toward Nation B. The weightage will help to identify the significance of each trust perceptions.
Theorem 1.
Mass Weightage Assume that: 𝑥 ∈ ℤ, 𝑥 ≥ 1;
𝒞 = |𝑅𝐸𝐿𝐴𝑇𝐼𝑂𝑁| 𝑜𝑟 𝑐𝑎𝑟𝑑𝑖𝑛𝑎𝑙𝑖𝑡𝑦(𝑅𝐸𝐿𝐴𝑇𝐼𝑂𝑁)
Mapped matrix:
𝑹𝑬𝑳𝑨𝑻𝑰𝑶𝑵 × (𝟏 ≤ 𝑿 ≤ 𝓒) 𝑻 [𝒇 𝒏 𝒉] × [𝟏𝟐𝟑] 𝑠. 𝑡. (𝑓 ↦ 1), (𝑛 ↦ 2) 𝑎𝑛𝑑 (ℎ ↦ 3). 𝒲 𝑀𝑎𝑠𝑠 = ∑ 𝒲 𝑥𝑥=1𝑥≤𝒞 = 1, 𝒲 𝑥 ∈ ℝ, 0 ≤ 𝒲 𝑥 ≤ 1 𝑒. 𝑔. 𝑤ℎ𝑒𝑛 𝑥 = 1 , 𝑡ℎ𝑒𝑛 𝒲 𝑥 𝑖𝑠 𝑓𝑜𝑟 𝑓𝑟𝑖𝑒𝑛𝑑𝑙𝑦 ′ 𝑠 𝑤𝑒𝑖𝑔ℎ𝑡𝑎𝑔𝑒 Remark 4.1
One may choose to use a priori probability to evaluate (assign value) for every 𝓦 𝐱 . Given that a cardinality is equal to three, then each 𝓦 𝒙 is equal to . One may also to use a different value of 𝓦 𝒙 that is based on the number of trust properties 𝓟 as mentioned in Theorem 3. For large numbers of the properties 𝓟 for a given 𝓦 𝒙 , the 𝓦 𝒙 should be increased to represent large samples of the properties 𝓟 . However, the value of 𝓦 𝒙 is subjective to an observer. Definition 5 . Scalar is used to determine an interval scale for international nation relations that are either friendly, neutral, or hostile.
Theorem 2.
Mass Scalar Assume that: 𝒮 = ℎ ′ 𝑠 𝑠𝑐𝑎𝑙𝑎𝑟 𝑠𝑖𝑔𝑛 = −1 𝒮 = 𝑛′𝑠 𝑠𝑐𝑎𝑙𝑎𝑟 𝑠𝑖𝑔𝑛 = +1 𝒮 = 𝑓 ′ 𝑠 𝑠𝑐𝑎𝑙𝑎𝑟 𝑠𝑖𝑔𝑛 = +1 * One may choose scalar signs: either +ve or –ve) . 𝒮 𝑀𝑎𝑠𝑠 = ∑ |𝒮 𝑥 . 𝒲 𝑥 | 𝑥=1𝑥≤𝒞 = 1 Lemma 2.1.
Lower bound, middle bound and upper bound in Mass Scalar (interval scale). 𝒮 𝐿𝑜𝑤𝑒𝑟 = 𝒮 . 𝒲 𝒮 𝑢𝑝𝑝𝑒𝑟 = ∑ 𝒮 𝑥 . 𝒲 𝑥𝑥=2𝑥≤𝒞 𝒮 𝐿𝑜𝑤𝑒𝑟 + 𝒲 ≤ 𝒮 𝑚𝑖𝑑𝑑𝑙𝑒 ≤ (𝒮 𝑢𝑝𝑝𝑒𝑟 − 𝒮 𝒞 . 𝒲 𝒞 ) Definition 6.
Trust perception is a collection of trust properties or elements that are used in determining a trust alignment for ℛ 𝐴,𝐵 . Remark 6.1
Trust relations ℛ 𝐴,𝐵 will have the collection of trust properties 𝒫 for each trust perceptions (e.g. 𝑓 𝒫 , 𝑛 𝑝 , ℎ 𝑝 ). Each trust property 𝒫 𝑥 can be mapped into nominal data with values such as military 0.2, politic 0.3, trade 0.1, spying 0.05, etc. Theorem 3.
Mass Trust Properties Assume that: 𝒫 𝑥 ∈ ℝ, 𝑠. 𝑡. 0 ≤ 𝒫 𝑥 ≤ 1; Let cardinalities: 𝑖, 𝑗, 𝑘 𝑖 = |𝑓𝑟𝑖𝑒𝑛𝑑𝑙𝑦 𝑝𝑟𝑜𝑝𝑒𝑟𝑡𝑖𝑒𝑠| 𝑗 = |𝑛𝑒𝑢𝑡𝑟𝑎𝑙 𝑝𝑟𝑜𝑝𝑒𝑟𝑡𝑖𝑒𝑠| 𝑘 = |ℎ𝑜𝑠𝑡𝑖𝑙𝑒 𝑝𝑟𝑜𝑝𝑒𝑟𝑡𝑖𝑒𝑠| ℎ 𝒫 𝑀𝑎𝑠𝑠 = ∑ 𝒫 𝑥𝑥=1𝑥=𝑘 ≤ 1 We choose to use a negative sign for a hostile and positive sign for a neutral and friendly relations. In common sense, the negative sign may suitable to be used for the hostile relation. 𝑛 𝒫 𝑀𝑎𝑠𝑠 = ∑ 𝒫 𝑥𝑥=1𝑥=𝑗 ≤ 1 𝑓 𝒫 𝑀𝑎𝑠𝑠 = ∑ 𝒫 𝑥𝑥=1𝑥=𝑖 ≤ 1
Definition 7.
Trust relations for ℛ 𝐴,𝐵 is a product of Mass Trust Perception 𝓣 𝑴𝒂𝒔𝒔 . The Mass Trust Perception is a point that resides in a relative distance between a lower bound and upper bound of Mass Scalar. To determine the trust relations for the ℛ 𝐴,𝐵 , i.e. either friendly, neutral or hostile: If the point falls into less than middle bound, it is a hostile relation; If the point falls into greater than middle bound, it is a friendly relation; If the point falls into the middle bound, it is a neutral relation.
Remark 7.1
Theorems 1 through 4 rely on three major conditions of relations between nations (hostile, neutral and friendly). One may define more than triple conditions to implement granularity and fuzziness in the relations.
Remark 7.2
One should not modify in order to implement an additional condition of relations because it will increase difficulties in properties 𝒫 classification and nominal data (value assignment). Our suggestion for more than the triple scalar relations is to directly map the Mass Trust Perception’s value in Theorem 4 into the septuple scalar. One must define a lower bound and an upper bound for each new relation element. The new relation element is a subset of the existing triple (e.g. Near-Hostile ⊂ Hostile).
Theorem 4.
Mass Trust Perception Mapped matrix
𝑴𝒂𝒔𝒔 𝑻𝒓𝒖𝒔𝒕 𝑷𝒓𝒐𝒑𝒆𝒓𝒕𝒊𝒆𝒔 × 𝐒𝐜𝐚𝐥𝐚𝐫 𝒯 𝑀𝑎𝑠𝑠 = ∑( [𝑓 𝒫 𝑀𝑎𝑠𝑠 𝑛 𝒫 𝑀𝑎𝑠𝑠 ℎ 𝒫 𝑀𝑎𝑠𝑠 ] × [𝒮 . 𝒲 𝒮 . 𝒲 𝒮 . 𝒲 ] ) ℛ 𝐴,𝐵 (𝒯 𝑀𝑎𝑠𝑠 ) = { ℎ𝑜𝑠𝑡𝑖𝑙𝑒, 𝒯
𝑀𝑎𝑠𝑠 < 𝒮
𝐿𝑜𝑤𝑒𝑟 − 𝒲 𝑛𝑒𝑢𝑡𝑟𝑎𝑙, 𝒯 𝑀𝑎𝑠𝑠 = 𝒮 𝑚𝑖𝑑𝑑𝑙𝑒 𝑓𝑟𝑖𝑒𝑛𝑑𝑙𝑦, 𝒯
𝑀𝑎𝑠𝑠 > 𝒮 𝑢𝑝𝑝𝑒𝑟 − 𝒮 𝒞 . 𝒲 𝒞 Lemma 4.1.
Strength of Mass Trust Perception 𝒯 𝑀𝑎𝑠𝑠 𝑆𝑡𝑟𝑒𝑛𝑔𝑡ℎ = ∑( [𝑓 𝒫 𝑀𝑎𝑠𝑠 𝑛 𝒫 𝑀𝑎𝑠𝑠 ℎ 𝒫 𝑀𝑎𝑠𝑠 ] × [𝒲 𝒲 𝒲 ] ) When 𝒯 𝑀𝑎𝑠𝑠 𝑆𝑡𝑟𝑒𝑛𝑔𝑡ℎ is near to 1, 𝒯 𝑀𝑎𝑠𝑠 may represent many contradiction of opinions between hostile and friendly relations. This may happen if it involves a long duration of sampling (or observation) of international relations between two nations. If the 𝒯 𝑀𝑎𝑠𝑠 is a product of 20 years observation of the international relations between two nations, it may consist of a year of war, a year of military allies, a year of politics disagreement, a year of economy sanctions, etc. If the 𝒯 𝑀𝑎𝑠𝑠 is a product of shorter years observation, the contradiction of opinions may occur when a nation leader or ruling party was changed due to election, revolution, installation of puppet leader as a post-war outcome, etc. When 𝒯 𝑀𝑎𝑠𝑠 𝑆𝑡𝑟𝑒𝑛𝑔𝑡ℎ is near to 0.5 (or middle), 𝒯 𝑀𝑎𝑠𝑠 may represent a fair opinion that either hostile or friendly relations. If the 𝒯 𝑀𝑎𝑠𝑠 is a product of observation for many years, it may represent consistent international relations during that duration. When 𝒯 𝑀𝑎𝑠𝑠 𝑆𝑡𝑟𝑒𝑛𝑔𝑡ℎ is near to 𝑛 𝒫 𝑀𝑎𝑠𝑠 , 𝒯 𝑀𝑎𝑠𝑠 represents a bias to a neutral. If the 𝒯 𝑀𝑎𝑠𝑠 is a product of observation for many years, it may represent a firm of neutral relations at that moment. When 𝒯 𝑀𝑎𝑠𝑠 𝑆𝑡𝑟𝑒𝑛𝑔𝑡ℎ and 𝑛 𝒫 𝑀𝑎𝑠𝑠 are identical in a positive value, it indicates that there is no hostile property in the calculation (or observation). Fig. 1.
Summary Trust Algebra for Relation Computation
Figure 1 shows a summary of trust computation between two nations. To initialize the trust computation, one should identify trust properties 𝒫 and mass weightage for each relation (e.g. Table 1). Let’s observe the following example: Table 1: Trust Computation Example
Relation Hostile Neutral Friendly Properties 𝒫 𝑥 𝒫 = 0.5 𝒫 = 0.3 𝒫 = 0 𝒫 = 0.15 𝒫 = 0.05 𝒫 = 0 𝒫 = 0.5 𝒫 = 0 𝒫 = 0 𝒫 = 0.1 𝒫 =0.05 𝒫 =0 𝒫 =0 𝒫 = 0 𝒫 = 0 𝒫 = 0.1 Weightage 𝒲 𝑀𝑎𝑠𝑠 𝒮 𝑥 𝒮 𝑖𝑠 − 𝒮 𝑖𝑠 + 𝒮 𝑖𝑠 + ℎ 𝒫 𝑀𝑎𝑠𝑠 = 0.9 = 0.5 + 0.2 + 0.15 + 0.05 𝑛 𝒫 𝑀𝑎𝑠𝑠 = 0.6 = 0.5 + 0.1 𝑓 𝒫 𝑀𝑎𝑠𝑠 = 0.15 = 0.05 + 0.1 𝒮 𝑀𝑎𝑠𝑠 = 1 = 0.45 + 0.10 + 0.45 𝒮 𝐿𝑜𝑤𝑒𝑟 = −0.45 𝒮 𝑢𝑝𝑝𝑒𝑟 = 0.55 = 0.10 + 0.45 −0.45 + 0.45 ≤ 𝒮 𝑚𝑖𝑑𝑑𝑙𝑒 ≤ (0.55 − 0.45)
0 ≤ 𝒮 𝑚𝑖𝑑𝑑𝑙𝑒 ≤ 0.10 𝒯 𝑀𝑎𝑠𝑠 = ∑( [0.9 0.6 0.15] × [−. 0.45+.0.10+. 0.45] ) 𝒯 𝑀𝑎𝑠𝑠 = −0.2775 = ∑( [−0.405 +0.06 +0.0675]) ℛ 𝐴,𝐵 (−0.2775) = { 𝒉𝒐𝒔𝒕𝒊𝒍𝒆, 𝓣
𝑴𝒂𝒔𝒔 𝒊𝒔 < 𝑺_𝒎𝒊𝒅𝒅𝒍𝒆𝑛𝑒𝑢𝑡𝑟𝑎𝑙, 𝒯
𝑀𝑎𝑠𝑠 𝑖𝑠 𝑆_𝑚𝑖𝑑𝑑𝑙𝑒𝑓𝑟𝑖𝑒𝑛𝑑𝑙𝑦, 𝒯
𝑀𝑎𝑠𝑠 𝑖𝑠 > 𝑆_𝑚𝑖𝑑𝑑𝑙𝑒 𝒯 𝑀𝑎𝑠𝑠 𝑆𝑡𝑒𝑛𝑔𝑡ℎ = 0.5325 = ∑( [0.9 0.6 0.15] × [ 0.450.10 0.45] ) Referring to Theorem 4 and Lemma 4.1, the given example has shown that a relation between Nation A and Nation B is hostile. The strength of the relation is near to 0.5 such that it represents a consistent hostile relation during the observation.
Case Study: International Relations
In this section, we explore international nation relations between the United States of America and Great Britain (USA–GB.
Properties
We have clustered events that may affect international nation relations as showed in Tables 2, 3 and 4. Clustering or grouping the related events for certain properties will reduce complexities for determining properties’ values and it will help to reduce the searching time of the whole data in public domains (e.g. Internet, news, etc.). If at least a single event is found to be related to the given properties, then the given properties will be included in a trust computation. It may not be strong enough as a solid evidence for the given properties, but it will help to enable the trust computation. The Dempster-Shafer’s theory of evidence may also be applied in event verifications. However, it requires too much effort.
Table 2. Friendly (Positive) 𝓟 𝐱 Descriptions 𝒫 𝒫 𝒫 𝒫 𝒫 𝒫 Table 3. Neutral 𝓟 𝐱 Descriptions 𝒫 𝒫 𝒫 Table 4. Hostile (Negative) 𝓟 𝐱 Descriptions 𝒫 𝒫 𝒫 𝒫 𝒫 𝒫 Weightage
We chose to implement 40%:20%:40% as weightages for hostile, neutral and friendly relations. The weightage percentages were decided based on the number of properties for the given relations.
A Case Study: The USA and GBR (2001-2005)
The United States of America and Great Britain enjoy a long lasting of good international relations. The British-America (or Anglo-American) relation remains intact as close military allies since the World War II. Both nations also share various strategic information (e.g. UKUSA Agreement (NSA, 2013)).
Table 5. The USA and GBR ℛ 𝑈𝑆𝐴,𝐺𝐵
Hostile Neutral Friendly Properties 𝒫 𝑥 𝒫 = 0 𝒫 = 0 𝒫 = 0 𝒫 = 0 𝒫 = 0 𝒫 = 0 𝒫 = 0.25 𝒫 = 0.35 𝒫 = 0.40 𝒫 = 0.5 𝒫 = 0 𝒫 = 0.1 𝒫 = 0 𝒫 = 0.075 𝒫 = 0.025 Weightage 𝒲 𝑀𝑎𝑠𝑠 𝒮 𝑥 𝒮 𝑖𝑠 − 𝒮 𝑖𝑠 + 𝒮 𝑖𝑠 + ℎ 𝒫 𝑀𝑎𝑠𝑠 = 0 𝑛 𝒫 𝑀𝑎𝑠𝑠 = 1 𝑓 𝒫 𝑀𝑎𝑠𝑠 = 0.70 𝒮 𝑀𝑎𝑠𝑠 = 1 𝒮 𝐿𝑜𝑤𝑒𝑟 = −0.4 𝒮 𝑢𝑝𝑝𝑒𝑟 = 0.6 𝑚𝑖𝑑𝑑𝑙𝑒 ≤ 0.2 𝒯 𝑀𝑎𝑠𝑠 = 0.48 ℛ 𝑈𝑆𝐴,𝐺𝐵 (0.48) = { ℎ𝑜𝑠𝑡𝑖𝑙𝑒, 𝒯
𝑀𝑎𝑠𝑠 𝑖𝑠 < 𝒮 𝑚𝑖𝑑𝑑𝑙𝑒 𝑛𝑒𝑢𝑡𝑟𝑎𝑙, 𝒯
𝑀𝑎𝑠𝑠 𝑖𝑠 𝒮 𝑚𝑖𝑑𝑑𝑙𝑒 𝒇𝒓𝒊𝒆𝒏𝒅𝒍𝒚, 𝓣
𝑴𝒂𝒔𝒔 𝒊𝒔 > 𝓢 𝒎𝒊𝒅𝒅𝒍𝒆 𝒯 𝑀𝑎𝑠𝑠 𝑆𝑡𝑒𝑛𝑔𝑡ℎ = 0.48 ℛ 𝑈𝑆𝐴,𝐺𝐵 showed that the 𝒯 𝑀𝑎𝑠𝑠 and 𝒯 𝑀𝑎𝑠𝑠 𝑆𝑡𝑒𝑛𝑔𝑡ℎ are identical. When both variables are identical in a positive value, it indicates that there is no hostile properties in the observation. The strength of the relations is near to 0.5, which represents a consistent friendly relations in 2001 until 2005.
Results and Discussion
We have presented case studies for the trust algebra in Trust Algebra section. Both case studies discussed the international nation relations between USA-GB. The properties and weightages are subjective to the observers. In this work, the properties and weightages chosen by the authors are based on public information available in the literatures (refer to the Literature Review section) and the Internet . Based on limited information on the Internet, we have drawn tentative conclusions for ℛ 𝑈𝑆𝐴,𝐺𝐵 relations in 2001 until 2005. ℛ 𝑈𝑆𝐴,𝐺𝐵 𝑚𝑎𝑠𝑠 (0.48) showed positive relations, which fall into a friendly We do not obtain or use any material that may lead to actions of a cyber-crime, terrorism, spying or any other illegal activities. threshold. ℛ 𝑈𝑆𝐴,𝐺𝐵 𝑚𝑎𝑠𝑠 𝑠𝑡𝑟𝑒𝑛𝑔𝑡ℎ (0.48) is equal to ℛ 𝑈𝑆𝐴,𝐺𝐵 𝑚𝑎𝑠𝑠 , which implies that there are no hostile properties in the observation. The given tentative conclusions may change due to new evidence and new events that will be known by the observer in future.
Conclusion
In this work, we have modeled a trust algebra for international nation relations. The purpose of trust algebra method is to allow trust computations and trust modeling. Previously, there is no such a method to perform the trust computations for international relations which are subjective and unquantifiable. We have also presented the international nation relations between USA-GB as a case study to demonstrate the proposed method in a real-world scenario.