METHODOLOGY

Measures two independent dimensions: (1) the ratio of individual donors to corporate PAC money, and (2) top-donor concentration — what fraction of total fundraising comes from the top 10 donors. PAC dependency is operationalized following Stratmann (2005), [5] Stratmann 2005who found that PAC contributions are more strongly correlated with roll-call alignment than individual contributions. The donor concentration component applies the same intuition as HHI but at the donor level, following Bonica (2014) who demonstrated that donor composition is a strong predictor of legislative behavior. [1] Bonica 2014

Tracks whether a senator's voting record aligns with their stated campaign promises. Platform text is extracted from official senate.gov websites and analyzed to identify key commitments. Votes are then cross-referenced against those promises using semantic search to find relevant legislation. [2] Naurin 2011

A confidence penalty is applied when few promises are evaluable: if only 1 of 10 promises could be checked against votes, the score blends toward 50 (neutral) rather than being inflated by a single data point. This implements Bayesian shrinkage toward the prior. [19] Efron & Morris 1975

This metric also incorporates floor advocacy analysis — whether a senator actively speaks on the Senate floor about their promised issues, parsed from Congressional Record proceedings. This captures effort that voting records miss: in a gridlocked Senate, a senator may not get bills to a vote but can still demonstrate persistence through floor speeches. The floor advocacy component is weighted at 15% of the promise score, following research on legislative speech as a signal of commitment. [3] Martin 2011

Measures willingness to break with party leadership on votes that are not explained by constituent interests. We identify state-relevant policy areas by analyzing the senator's top donor industries (which serve as a proxy for the state's economic composition). Party-line votes on state-relevant issues are excluded from the independence penalty because they may reflect genuine constituent representation rather than blind party loyalty. [4] Carson et al. 2010

The score is adjusted by state partisan lean using Cook PVI as a proxy: a senator in a safe R+20 state voting with their party may be representing constituents, not following orders. Raw break rates are misleading without this contextual adjustment.

The score blends two components: party independence (60%) — the rate of breaking with the party on non-state-relevant votes — and donor independence (40%) — whether votes appear free from donor influence, measured by lobbying match alignment and PAC funding levels. We follow the methodological caution of Ansolabehere et al. (2003) [18] Ansolabehere et al. 2003in interpreting donation-vote correlations: correlation does not prove causation. [5] Stratmann 2005

Evaluates how traceable and diverse a senator's funding sources are. The score blends donor traceability (50%) — the fraction of funding from itemized (>$200), disclosed sources versus anonymous small-dollar contributions — with industry diversity (50%), measured as the inverse Herfindahl-Hirschman Index (HHI) of industry donations. HHI is a standard concentration metric from industrial organization economics; [6] Rhoades 1993in this context, funding concentrated in a single industry suggests potential regulatory capture, while broad funding suggests diverse constituent support.

Measures how effective a senator is at advancing legislation. The score combines bill passage rates, cosponsorship network influence (PageRank), and ability to move bills through committee and floor consideration. Higher scores indicate senators who successfully shepherd bills into law and attract bipartisan cosponsorship.

Measures legislative influence using the PageRank algorithm [32] Brin & Page 1998applied to cosponsorship networks. When Senator A cosponsors Senator B's bill, that creates a directed link in the network. PageRank computes centrality: a senator whose bills attract many cosponsors — especially from other influential senators — receives a higher score. This mirrors GovTrack's leadership methodology. [33] Tauberer 2012

The algorithm uses power iteration with a damping factor of 0.85 and converges in ~50 iterations. Raw PageRank values are rescaled to [0, 1] using a logarithmic transformation to compress the heavy-tailed distribution, then displayed as 0-100.

Computes a behavioral ideological position using Singular Value Decomposition (SVD) on the cosponsorship matrix, following Tauberer (2012). [33] Tauberer 2012The second singular vector (first is trivially related to overall activity) captures the primary ideological dimension — the axis along which senators most differ in who they cosponsor. This is analogous to DW-NOMINATE [20] Poole & Rosenthal 1985but derived from cosponsorship patterns rather than roll-call votes.

The ideology score is oriented so that lower values correspond to progressive positions and higher values to conservative positions, calibrated by checking the mean score of each party. It serves as a Bayesian prior for the partisan depth calculation: when a senator has few recorded votes, the ideology score regularizes the estimate; as vote data accumulates, the prior weight drops to zero. [19] Efron & Morris 1975

Combines the leadership and ideology scores into a human-readable label (e.g., "progressive Democratic leader" or "conservative Republican backbencher"). The label encodes three dimensions: ideological position (progressive/moderate/conservative), party affiliation, and influence tier (leader/rank-and-file/backbencher).

Assesses cabinet and advisor independence from corporate and lobbyist influence. Based on historical analysis of cabinet compositions — how many appointees came from industry versus public service backgrounds. Currently uses curated seed data; automated analysis is planned.

Measures the ratio of campaign promises to executive and legislative action. Based on historian assessments and promise-tracking analysis. Currently uses curated seed data for historical presidents.

Reflects approval trajectory and coalition retention. For modern presidents (Truman onward), this is grounded in Gallup average approval ratings. Pre-Gallup presidents are scored based on election margins and historian consensus.

Measures tangible economic outcomes: GDP growth and job creation. The pipeline fetches real employment data from the Bureau of Labor Statistics API (nonfarm payroll series CES0000000001) and calculates net jobs created during each term. GDP growth averages are seeded from BEA National Income and Product Accounts tables. The formula weights GDP at 60% and job creation at 40%, normalized against post-WWII historical averages.

Evaluates administrative execution quality. The pipeline fetches executive order counts from the Federal Register API (federalregister.gov) for each presidential term. The score blends court success rate (40%), cabinet stability (30%), and EO activity rate (30%), where a moderate rate of executive action scores higher than extremes in either direction.

Measures how well executive agency actions align with stated presidential priorities. Evaluates whether federal agencies pursue the policy agenda the president campaigned on, based on regulatory actions and executive directives. Currently uses curated seed data; automated regulatory tracking is planned.

NEWS ANALYSIS PIPELINE

RSS feeds from editorially independent, low-bias news sources (AP News, NPR Politics, Reuters, PBS NewsHour) are parsed hourly. Each article is filtered for U.S. policy relevance using embedding cosine similarity against policy area prototypes — the same sentence-transformer model used throughout the platform. Articles that pass the relevance threshold are clustered by semantic similarity to group coverage of the same story across sources.

TRENDING TOPIC INTEGRATION

Clusters are ranked using a weighted combination of coverage breadth (40%) — how many independent sources cover the story — and trending relevance (60%) — whether the topic aligns with what the public is actively discussing. Trending signals are drawn from Google Trends and policy-relevant Reddit communities, cross-referenced with the news clusters via embedding similarity.

NON-PARTISAN SUMMARIZATION

The top-ranked issues are summarized by the LLM with explicit instructions to present objective facts, avoid opinion or editorial framing, and recommend actions that do not assume which side of an issue the reader supports. Recommended actions include contacting representatives, attending public hearings, and reviewing primary source documents — not advocating for or against any policy position.

CROSS-REFERENCING

When a ranked politician is involved in a trending issue, the Action Center links directly to their scorecard. Related government documents from the Explore database are matched using semantic search. Source articles include direct links to the original reporting.

GOVERNMENT ACTIVITY TABS

Dedicated tabs for all three branches of government — Legislative (Senate and House), Executive, and Judicial — display the most recent government documents: floor speeches, executive orders, proposed rules, court opinions, and notices, pulled directly from the Explore database.

NATIONAL MONITORS

When an issue persists in the news across multiple days, the system automatically creates a National Monitor — a dedicated tracking page for that ongoing concern. Monitors build a sourced timeline of developments, detect when separate news stories are facets of the same underlying event using embedding similarity, and merge duplicate monitors automatically. Monitors transition to "watching" status when coverage subsides and reactivate when new developments appear.

YEAR-IN-REVIEW TIMELINE

Each day's top issue is permanently recorded in a timeline that accumulates throughout the calendar year. The Timeline tab provides a month-by-month chronological view of what mattered most, with top policy themes calculated for each month and the year as a whole. At year's end, this becomes a complete "Year in Review" of the issues that shaped civic life.

ELECTIONS TAB

The Elections tab displays upcoming election dates, Senate races with incumbent scores linked to their scorecards, and an interactive U.S. map for selecting states. State-specific information helps users understand their local races in the context of national trends.

INTERACTIVE GLOBAL NEWS MAP

The World tab features a 3D interactive globe that visualizes U.S.-related international news coverage. Countries mentioned in current news feeds are highlighted with points scaled by article count. Clicking a country scrolls to recent headlines about U.S. relations with that nation, linking to the original source articles.

HOW IT WORKS

We implement a nearest-centroid classifier (Rocchio 1971) [22] Manning, Raghavan & Schütze 2008in sentence-embedding space. Each party's known platform positions on each policy area (taxes, healthcare, environment, etc.) are embedded as centroids using the same sentence-transformer model used throughout the pipeline. Bill text is then embedded and compared to both party centroids via cosine similarity.

The stance direction (pro/anti) disambiguates cases where both parties have positions on the same topic: a "pro" environment bill (strengthen EPA enforcement) aligns with the Democratic platform, while an "anti" environment bill (roll back regulations) aligns with the Republican platform. This encodes the saliency-plus-direction model from manifesto research. [23] Laver & Garry 2000

TWO-SIGNAL FUSION

Content analysis is the primary signal for party alignment. Vote tallies from roll-call data serve as a secondary refinement. When both agree, confidence is high. When they disagree, content wins unless the vote data shows a clear party-line split (which is itself informative — the bill was important enough to whip). This follows Snyder & Groseclose (2000) who demonstrated that vote outcomes reflect party discipline as much as ideology. [24] Snyder & Groseclose 2000

ADAPTIVE LEARNING

Platform position descriptions are seed prototypes bootstrapped from published party platforms. As the pipeline processes bills, sponsor party data from Congress.gov serves as supervised ground truth — bills sponsored by a single party are labeled examples that refine the classifier over time. This follows the self-training paradigm. [25] Yarowsky 1995

BILL POLICY AREA CLASSIFICATION

Bills and votes are classified into 15 policy areas (healthcare, defense, energy, etc.) using a tiered adaptive strategy:

• 1.Learning store exact match — bills classified in prior pipeline runs are recalled instantly by ID. This is the experience replay pattern. [10] Lin 1992
• 2.kNN against reference corpus — the k=7 most similar previously-classified bills in ChromaDB are retrieved and the policy area is assigned by similarity-weighted majority vote. [9] Cover & Hart 1967This is retrieval-augmented classification: the reference corpus grows with each pipeline run, improving accuracy over time. [26] Lewis et al. 2020
• 3.Embedding similarity against policy descriptions — cosine similarity between bill text embeddings and pre-computed policy area description embeddings. This is the cold-start fallback using nearest-centroid classification. [7] Reimers & Gurevych 2019

The policy taxonomy is based on the Congressional Research Service (CRS) policy area scheme used by Congress.gov. The approach follows the text-as-data paradigm reviewed in Grimmer & Stewart (2013). [27] Grimmer & Stewart 2013Stance derivation (pro/anti/neutral) uses embedding cosine similarity against direction prototypes — the bill text is compared to semantic signatures of supportive, restrictive, and reform-oriented legislative language, following the Comparative Agendas Project coding tradition. [28] Baumgartner & Jones 1993Zero LLM calls are used for bill classification.

DONOR AND INDUSTRY CLASSIFICATION

Donor classification uses a five-tier strategy:

• 1.FEC metadata — structured fields from the Federal Election Commission API encode committee type and designation codes, providing ground-truth classification for PACs vs. individual donors.
• 2.Semantic detection — embedding cosine similarity against category prototypes replaces ~200 lines of hardcoded string patterns. This generalizes to unseen entities because distributed representations capture semantic meaning. [29] Bengio et al. 2003
• 3.Learning store lookup — previously classified entities are recalled instantly by name.
• 4.Embedding cosine similarity — donor names are compared against pre-computed industry description embeddings. Industry descriptions include exemplar company names as anchoring tokens, following the zero-shot classification setup. [30] Yin, Hay & Roth 2019
• 5.k-Nearest Neighbor (kNN) — remaining unclassified donors are classified by the k=7 most similar already-labeled entities using distance-weighted majority voting. [9] Cover & Hart 1967This mirrors prototypical networks for few-shot learning [14] Snell et al. 2017where classification is performed by comparing query embeddings to accumulated real examples.

The kNN approach was chosen over LLM-based classification after empirical testing showed the LLM hallucinated invalid categories (producing labels like "SPORTS" or "RESTAURANT" outside the valid taxonomy) and was orders of magnitude slower. The kNN classifier processes ~5,000 donors in under 5 seconds versus 40+ minutes for the LLM, with more consistent results.

LEARNING STORE AND ADAPTIVE CLASSIFICATION

All classifications are persisted in a learning store (SQLite table) that functions as an evolving knowledge base. On subsequent pipeline runs, previously classified entities are retrieved instantly without recomputation. This is analogous to experience replay in reinforcement learning [10] Lin 1992 — past decisions inform future ones, improving both speed and accuracy over time.

The learning store also feeds into the self-training loop [25] Yarowsky 1995 — high-confidence classifications from prior runs become labeled examples for kNN and reference corpus retrieval in future runs. The system literally gets better each time the pipeline runs.

To prevent stale data from persisting when analysis algorithms are updated, the pipeline implements version-aware artifact management. At the start of each run, a SHA-256 fingerprint of all analysis source files is compared to the stored hash from the previous run. If the code is unchanged, all learning data is preserved to promote self-training. If the code has changed, stale artifacts (LLM results, learned classifications, kNN reference corpus) are automatically cleared so updated algorithms start fresh. The API cache (raw data from government APIs) is never cleared.

SEMANTIC SEARCH (EXPLORE)

The Explore feature uses dense passage retrieval [11] Karpukhin et al. 2020 to enable free-text search over government documents (floor speeches, executive orders, bills). Documents are chunked, embedded with all-MiniLM-L6-v2, and stored in ChromaDB for approximate nearest-neighbor retrieval. This outperforms keyword search (BM25) for conceptual queries like "climate policy" where exact term overlap is low.

EMBEDDING MODEL (CLASSIFICATION + SEARCH)

all-MiniLM-L6-v2 (22M parameters) [8] Wang et al. 2020handles all classification tasks: bill policy areas, donor industries, party alignment, motion types, and semantic search retrieval. Sentence-transformers produce dense vector representations where cosine similarity correlates with semantic similarity [7] Reimers & Gurevych 2019 — making them ideal for classification-by-comparison tasks where category definitions exist.

LLM (NARRATIVE SYNTHESIS)

Qwen 2.5 1.5B via llama.cpp [16] Gerganov 2023 handles tasks requiring natural language understanding and multi-step reasoning:

WHAT AI DOES NOT DO

WHY THESE TECHNIQUES WERE CHOSEN

We follow a strict hierarchy: structured metadata first, then embedding similarity, then kNN, then LLM — reserving each more expensive technique only for tasks the cheaper ones cannot handle. The pipeline contains zero hardcoded keyword lists, regex patterns, or string-matching heuristics for classification decisions. Every classification is made mathematically via embedding cosine similarity against natural-language prototypes. [12] Jurafsky & Martin 2023

Embeddings (not LLM) for classification: sentence embeddings excel at text classification tasks when labeled examples or category descriptions exist. They are deterministic, fast, and avoid the hallucination risks inherent in generative models. [13] Minaee et al. 2021The kNN classifier further leverages accumulated labeled data as a growing reference set — a well-established approach in few-shot and semi-supervised learning settings. [14] Snell et al. 2017

Content analysis (not votes) for party alignment: roll-call votes confound ideology with legislative strategy. Analyzing what a bill does relative to published party platforms recovers ideological alignment more accurately, following the manifesto analysis tradition. [31] Laver, Benoit & Garry 2003

LLM for narrative synthesis: tasks like promise-vote cross-referencing and PAC identification require world knowledge and multi-step reasoning that embeddings alone cannot provide. These are inherently generative tasks suited to language models. [15] Wei et al. 2022

MODEL AND ARCHITECTURE

The inference model is Qwen 2.5 1.5B, a compact open-weight language model running natively via llama.cpp [16] Gerganov 2023 compiled with ARM-specific optimizations (cortex-a76, dot-product, fp16). This provides ~3x faster inference compared to containerized runtimes, generating ~8 tokens/second on the Raspberry Pi 5 CPU. Results are cached in a local database so each unique analysis is computed at most once.

The embedding model is all-MiniLM-L6-v2 [8] Wang et al. 2020, a 22M-parameter sentence transformer. It handles all classification (bills, donors, industries, party alignment), semantic search, and nearest-neighbor retrieval. Both models run entirely on-device with no external API calls.