Positional Scoring Matching Rules

• Positional Scoring Matching Rules is a framework that assigns parameterized scores to ranking positions, aggregating individual scores to determine winners from full or partial queries.
• The framework employs binomial-basis vectors and query complexity analysis to characterize when scoring rules can be computed, revealing intrinsic limitations such as the need for full rankings in plurality scenarios.
• Applications span voting protocols, sports tournaments, string matching algorithms, and positional games, with both geometric and optimal scoring families offering practical trade-offs and equilibrium insights.

A positional scoring matching rule is a framework wherein alternatives, candidates, or patterns are ranked or matched according to positionally-weighted scores defined by a parameterized scoring vector, and outcomes (rankings, winners, or matches) are determined by aggregating these positional scores over multiple contests, queries, or data alignments. These rules underpin diverse domains including voting protocols, rank aggregation, competitive tournament systems, and even exact string matching algorithms.

1. Fundamental Definitions and Mathematical Frameworks

Positional scoring rules assign to each position $j$ in an ordering of $m$ alternatives a real score $s_j$, with $s_1 \geq s_2 \geq \cdots \geq s_m$. When aggregating $n$ full rankings $\sigma_1, \dots, \sigma_n$ (e.g., voter ballots), the total score of candidate $c$ is

$$\mathrm{sc}_s^\pi(c) = \mathbb{E}_{\sigma \sim \pi} [s_{\text{position of } c \text{ in } \sigma}]$$

where $\pi$ is a profile-distribution over all $m!$ orderings (Halpern et al., 2024).

In tournaments and sports, these vectors define how many points are awarded for each contest position over a season of results, and the order of total points determines the champion (Csató, 2021). In string matching, each relative position in the pattern is assigned an advancement score based on letter frequencies, yielding a "positional scoring" of pattern shifts (Cantone et al., 2010).

2. Characterization of Computability from Incomplete Information

The central decision question is: for which scoring vectors $m$0 can one -- using only partial information (queries of $m$1 alternatives at a time) -- reliably determine the aggregate winner?

Main Characterization (Halpern et al., 2024):

• For every $m$2 and $m$3, a scoring rule $m$4 is computable with $m$5-queries iff $m$6 lies in the subspace $m$7 spanned by the binomial-basis vectors

$m$8

Equivalently, only those scoring rules expressible as $m$9 can be computed from responses to all possible $s_j$0-element subset queries. Plurality (the $s_j$1 vector) is never in $s_j$2 for $s_j$3, and thus requires full rankings for information-theoretic computation.

A swap construction shows that if $s_j$4, then there exist $s_j$5-indistinguishable profiles with different $s_j$6-winners, signifying an intrinsic limitation.

3. Algorithmic and Query Complexity Analysis

Deterministic Query Complexity (Halpern et al., 2024):

• Let $s_j$7 be the smallest $s_j$8 with $s_j$9.
• The minimal number of $s_1 \geq s_2 \geq \cdots \geq s_m$0-queries required is

$s_1 \geq s_2 \geq \cdots \geq s_m$1

where $s_1 \geq s_2 \geq \cdots \geq s_m$2 is the size of the minimal $s_1 \geq s_2 \geq \cdots \geq s_m$3-set covering system for all $s_1 \geq s_2 \geq \cdots \geq s_m$4-subsets.

• The tight deterministic query complexity is $s_1 \geq s_2 \geq \cdots \geq s_m$5.

Randomized Algorithms (Halpern et al., 2024):

• If a randomized algorithm uses $s_1 \geq s_2 \geq \cdots \geq s_m$6 queries, the success probability $s_1 \geq s_2 \geq \cdots \geq s_m$7 is bounded by

$s_1 \geq s_2 \geq \cdots \geq s_m$8

For constant $s_1 \geq s_2 \geq \cdots \geq s_m$9, realizing any fixed improvement over uniform random guessing requires $n$0 queries.

Special Case: $n$1, $n$2 (Borda via pairwise queries):

• One random pair query yields success probability at most $n$3; two pair queries give at most $n$4.

4. Rule Selection: Axiomatic and Empirical Considerations

Geometric and Optimal Scoring Families (Kondratev et al., 2019, Csató, 2021):

• The "geometric" family arises by requiring independence of unanimous winners/losers: $n$5
  • $n$6: Plurality $n$7
  • $n$8: Borda $n$9
  • $\sigma_1, \dots, \sigma_n$0: Antiplurality $\sigma_1, \dots, \sigma_n$1.
• The "optimal" score vector for aggregated ranking given i.i.d. symmetric utilities $\sigma_1, \dots, \sigma_n$2 is

$\sigma_1, \dots, \sigma_n$3

where $\sigma_1, \dots, \sigma_n$4 is the $\sigma_1, \dots, \sigma_n$5-th order statistic (Kondratev et al., 2019).

Empirical Fit and Tradeoffs:

• Empirical analysis in contexts such as F1 racing reveals a Pareto frontier between minimizing early championship clinch probability and the risk of "winless" champions. Both historical and geometric (p ≈ 1.3–1.6) rules can achieve near-optimal trade-offs (Csató, 2021, Kondratev et al., 2019).
• In sports where time/score distributions are nearly uniform, a geometric rule with suitable $\sigma_1, \dots, \sigma_n$6 closely matches the optimal ordinal rule.

5. Applications Beyond Voting: String Matching and Games

String Matching (Cantone et al., 2010):

• The "worst-character rule" for pattern matching algorithms is a positional scoring matching rule: for each position $\sigma_1, \dots, \sigma_n$7 in the pattern, compute the expected advancement $\sigma_1, \dots, \sigma_n$8 for inspecting at $\sigma_1, \dots, \sigma_n$9, then select the $c$0 maximizing $c$1. This generalizes Horspool and Quick-Search heuristics, unifying classic pattern matching under the positional scoring matching umbrella.

Scoring Positional Games (Bagan et al., 2022):

• In 2-player positional games on graphs, players alternately claim vertices and score the number of fully controlled edges. In the Maker-Maker convention, differences in edge counts correspond to positional scoring under optimal adversarial selection. Computing optimal scores is PSPACE-complete in the Maker-Breaker setting but linear-time solvable in the Maker-Maker case.

6. Theoretical Implications for Stability and Equilibrium

Electoral Equilibria and Rule Shape (Cahan et al., 2013):

• The structure of the scoring vector $c$2 determines the existence and nature of Nash equilibria in electoral competition:
  • Strictly convex rules (rapid drop-off in scores) preclude nonconvergent equilibria (NCNE).
  • Borda-type or "block" rules enable multipositional NCNE, facilitating clustering.
  • The intermediate parameter $c$3 predicts equilibrium regimes:
  • Only convergent equilibria for $c$4 (e.g., Borda, antiplurality).
  • Multipositional or bi-positional equilibria for suitable best-rewarding rules.

7. Optimization and Learning of Scoring Rules

Learning Positional Scoring Rules (Caragiannis et al., 2016):

• When only partial rankings and weak supervision (pairwise constraints) are available, OptPSR seeks a scoring vector maximizing consistency with known constraints. Exact computation is polynomial-time for fixed $c$5 but NP-hard in general, with a tight hardness of $c$6 factor.
• Approximate methods, including BestApproval (selecting the best $c$7-approval rule) guarantee a $c$8-fraction of optimum; more refined schemes (ApxPSR$c$9) interpolate between speed and quality.
• Experimental evidence shows that simple scoring vectors (e.g., Harmonic) frequently recover nearly all ground-truth relations in practical rank aggregation.

---

Summary Table: Main Families and Their Properties

Scoring Vector Family	Defining Property	Key Example
Geometric ($$\mathrm{sc}_s^\pi(c) = \mathbb{E}_{\sigma \sim \pi} [s_{\text{position of } c \text{ in } \sigma}]$$0)	IUL & IUW axioms; $$\mathrm{sc}_s^\pi(c) = \mathbb{E}_{\sigma \sim \pi} [s_{\text{position of } c \text{ in } \sigma}]$$1	Borda ($$\mathrm{sc}_s^\pi(c) = \mathbb{E}_{\sigma \sim \pi} [s_{\text{position of } c \text{ in } \sigma}]$$2), Plurality ($$\mathrm{sc}_s^\pi(c) = \mathbb{E}_{\sigma \sim \pi} [s_{\text{position of } c \text{ in } \sigma}]$$3)
Optimal (for $$\mathrm{sc}_s^\pi(c) = \mathbb{E}_{\sigma \sim \pi} [s_{\text{position of } c \text{ in } \sigma}]$$4)	Aggregates via expected utilities	$$\mathrm{sc}_s^\pi(c) = \mathbb{E}_{\sigma \sim \pi} [s_{\text{position of } c \text{ in } \sigma}]$$5
$$\mathrm{sc}_s^\pi(c) = \mathbb{E}_{\sigma \sim \pi} [s_{\text{position of } c \text{ in } \sigma}]$$6-Query Compatible $$\mathrm{sc}_s^\pi(c) = \mathbb{E}_{\sigma \sim \pi} [s_{\text{position of } c \text{ in } \sigma}]$$7	Binomial-basis span	Computable from $$\mathrm{sc}_s^\pi(c) = \mathbb{E}_{\sigma \sim \pi} [s_{\text{position of } c \text{ in } \sigma}]$$8-subsets

---

References: (Halpern et al., 2024, Csató, 2021, Kondratev et al., 2019, Caragiannis et al., 2016, Cahan et al., 2013, Bagan et al., 2022, Cantone et al., 2010).