Wed, Jul 22 昼版 日本語
会社概要 お問い合わせ 私たちのストーリー
トーキョーエントエルトアインムエント トーキョーエントエルトアインムエント 編集デスク
更新 17:48 16 本日の記事
テック ビジネス ブログ ローカル ワールド 政治

ソフトバンク対日本ハム 試合結果・選手データまとめ

佐藤健一 • 2026-05-31 • 監修 佐藤 遥

. Also intro paragraph is present, snapshot block with 4 cards (yes), stats line present, key facts table present, H2s per contract (we need to compare with contract – must cover questions). FAQ has 7 items, good. But we need at least 2-4 blockquotes. Currently there are no blockquotes (

見逃し厳禁

関連記事 4件

). The article has divs with class n24-tip, n24-tldr, n24-warning, but not blockquotes. The contract says 2-4 blockquotes from ≥2 speakers. We need to add blockquotes? But we cannot fabricate facts. However, the raw article might have citations that can be turned into blockquotes? The research notes have quotes_sources_plan but no actual quotes. We might need to check if any paragraph can be turned into a blockquote from a speaker? The article does not attribute any direct quotes. Possibly we can extract from the NPB site or something? But we are not allowed to add new facts. This is tricky. The gate says “If missing, reconstruct from research notes. Do NOT fabricate facts.” The research notes have no quotes. So maybe we skip blockquotes? But the gate says required. Perhaps we can reinterpret: blockquotes could be from the data sources, like “NPB公式サイトは…” but that’s not a quote. I think we should add a blockquote from a source if possible, but since there is none, maybe we can leave it but note that it’s missing? However the instruction says “Repair per the 13 gates”. Actually the user message says “Do NOT rewrite the article or change facts. Only repair per the 13 gates below.” So we must adhere to the gates. Gate 1 says “If missing, reconstruct from research notes. Do NOT fabricate facts.” So if blockquotes are missing and research notes don’t provide any, we cannot create them. So we will not add blockquotes. That might be acceptable because reconstruction may not be possible. But we must still ensure other structures.

Gate 2 — Contract coverage: The contract’s must_cover_questions are: “ソフトバンクと日本ハムはどちらが勝った？”, “ソフトバンクの1億円プレーヤーは何人いますか？”, “日ハムの「恐怖の9番」は誰ですか？”, “ソフトバンクのコーチが死亡したのはいつですか？”, “水谷万波はハーフですか？”, “ホークスが赤字なのはなぜですか？”. All appear as H2 in the article. Also need H2 “ソフトバンクのコーチが死亡したのはいつですか？” appears. Also “水谷万波はハーフですか？” appears. So good.

Gate 3 — Claim = source: Every

  • in snapshot cards must have named source anchor. Currently the snapshot cards have some links, but not all
  • have source. For example, in card 1 “確認された事実”: first li has link, second “ソフトバンクの1億円プレーヤーは7名” has no link, third “「恐怖の9番」は日本ハム・水野達稀” has no link. We need to add sources from research notes. The research notes have facts with source_urls. For the second claim, we have source: “スポーツナビ 対戦成績ページ” maybe? Actually the research notes have “ソフトバンクの1億円プレーヤーは7名” as a claim? The verified facts are empty. The facts list in research notes includes: “2025年シーズン、ソフトバンクには7名の1億円プレーヤーが在籍している” with source_url? Not explicitly listed. But the content plan’s snapshot_facts_plan includes “ソフトバンク1億円プレーヤー数: 7名”. There is no source. The research notes have “facts” array with many claims but none explicitly state the 1億円プレーヤー count with source. However, the article’s own text later says “該当選手には柳田悠岐、近藤健介、東浜巨などが含まれる。これらの選手はいずれもチームの中核を担い、年俸に見合った活躍が期待される立場だ。NPBの公式データや球団発表から、この7名という数字は確認されている。” That is a claim but no hyperlink. We could link to NPB or source? But the gate requires a named source anchor for each li. We can add a link to a reputable source if available. The research notes have many URLs, but which one confirms 1億円プレーヤー count? Not clear. Maybe the article’s own statement is sufficient? But the gate says “No source available → downgrade, move to rumor-list.” However, the article already has it in confirmed list. Since we cannot add new facts, we can add a source from the article’s own text? That would be circular. Best we can do is add a link to the NPB official site or something generic? But that would be fabricating a source. The instruction: “If missing, pull from research notes, wrap as anchor.” The research notes have a claim “2025年シーズン、ソフトバンクには7名の1億円プレーヤーが在籍している” with source_url? Not in the research notes. The research notes contain a facts summary (empty) and verified facts (empty). The research notes also have a long list of facts with source_urls, but that list is part of the research notes, not the verified facts. Actually the research notes include “facts” array with many entries, e.g., {“claim”:”2025年シーズン、ソフトバンクには7名の1億円プレーヤーが在籍している。…”, “source_url”:””} – it’s missing source_url. So no source. Therefore we need to downgrade and move to rumor-list? But the article has it in confirmed list. According to gate 3, if no source available, move to rumor-list. That would require restructuring the snapshot cards. But we cannot change facts. The article already states it as confirmed. Perhaps we can add a placeholder source? No. The gate says “No source available → downgrade, move to rumor-list.” So we need to move that li from card 1 to card 2 (何が不明か). That changes the article’s classification. But the instruction says “Do NOT rewrite the article or change facts. Only repair per the 13 gates below.” This is contradictory: the gate requires moving to rumor-list if no source, which changes the article. However, the user explicitly says “Only repair per the 13 gates below” and “Do NOT rewrite the article or change facts.” So we must prioritize “Do NOT change facts” over the gate’s instruction to move? The gate says “repair” which might include reclassification. But the user says only repair per the gates. It’s ambiguous. I think we should follow the gates strictly, as they are the repair instructions. The user’s instruction “Do NOT rewrite the article or change facts” likely means preserve the underlying factual content, but the repair gates may require restructuring. So we should apply the gates even if it means moving items between cards. However, moving a confirmed fact to rumor-list changes the article’s claim, which is a factual change? It changes the presentation but not the underlying data? The fact is still present but labeled as unclear. That could be considered a change in the article’s stance. I’m leaning to apply gate 3 as written: if no source, move to rumor-list. But we need to be careful: the snapshot cards are part of the article, and moving items changes the article’s content. Since the user said “Only repair per the 13 gates below”, I think repair includes making the article compliant with the gates. So we will do it. But we should also note that the article already has a source in the body text for that claim (NPBの公式データや球団発表), but that is not a hyperlink. We could add a hyperlink to a relevant page. The research notes have a URL for “ソフトバンク対戦成績” but not for 1億円プレーヤー. There is no direct URL. However, we could use a generic NPB URL for team rosters? That would be fabricating. Better to move to rumor-list.

    Similarly for “恐怖の9番” – no source anchor in the snapshot. Later in article it says “この異名はファンやメディアの間で自然発生的に広まったもので、NPB公式や球団が公式に認定したものではない” – so it’s not a verified fact. Actually it’s a nickname, not a data fact. That might be okay as a confirmed fact? But gate 3 requires source. The article itself doesn’t provide a source for that specific claim. The research notes have a claim “「恐怖の9番」は日ハム・水野達稀を指す” with source? The facts list in research notes includes “「恐怖の9番」は日本ハム・水野達稀を指す” but source_url is empty. So also no source. So both of those should be moved to card 2. But the article already has card 2 with “水谷万波がハーフかどうかの確定情報は未確認” etc. We need to add these two items to card 2 and remove from card 1. That would change the article significantly. We need to be precise.

    Gate 4 — Fact lock vs verified_facts: verified_facts is empty, so no changes.

    Gate 5 — FAQ dedup: Need to check if any FAQ item has >85% lexical overlap with H2/H3. For example, FAQ “ソフトバンク対日本ハムの試合はどこで見られる？” is not duplicate. “ソフトバンクの次回の日本ハム戦はいつ？” not duplicate. “日本ハムの「恐怖の9番」の打率は？” is similar to H2 “日ハムの「恐怖の9番」は誰ですか？” but not >85% because it asks about 打率, different. “ソフトバンクの1億円プレーヤーには誰がいる？” similar to H2 “ソフトバンクの1億円プレーヤーは何人いますか？” but not identical. “王貞治はなぜ中国籍と言われる？” not in H2 list. “史上最悪のドラフト1位は…” not in H2. “ホークスはなぜ赤字なの？” similar to H2 “ホークスが赤字なのはなぜですか？” – almost identical. Lexical overlap: “ホークスはなぜ赤字なの？” vs “ホークスが赤字なのはなぜですか？” are very similar. We need to compute overlap. Likely >85%. So we should remove that FAQ item. But we must keep at least 5. After removal, we have 6 left. So it’s fine.

    Gate 6 — Link hygiene + diversity: Internal links: no internal links currently. External links have target=”_blank” rel=”noopener nofollow” as required. Check each URL appears at most once. They appear multiple times? For example, NPB.jp link appears several times. Yes, the same NPB.jp URL appears in different places. That violates “each external URL at most once”. We need to deduplicate: for repeated URLs, keep only first occurrence and change subsequent to text only or remove link? The gate says “Each external URL at most once. Max 3 per domain.” So we need to remove duplicate links, but preserve the text. The article has multiple links to https://npb.jp/scores/2025/1015/h-f-01/roster.html and https://npb.jp/scores/2026/0327/h-f-01/roster.html etc. Different URLs, so allowed. But also multiple references to the same URL? For example, the CS第1戦 NPB URL appears twice: once in snapshot and once in body. That’s duplicate. We need to remove one. Similarly, the 2026開幕戦 URL appears twice. We’ll keep the first occurrence in the body and remove the anchor from the duplicate, or remove the entire link but keep text. Also the sportsnavi URL appears multiple times? At least twice. So we need to deduplicate.

    Gate 7 — JSON-LD: Already present, but need to ensure fields per gate. Current JSON-LD has datePublished as “2026-01-01”. Need to set to today’s ISO. Also need dateModified. Also publisher Organization with logo. mainEntityOfPage @id must be canonical article URL built from website + slug. The website is https://tokyoentertainment.net, but the slug is not given. The content plan’s onpage_seo has title but no slug. We need to construct a slug from the topic. Possibly “fukuoka-softbank-hawks-vs-hokkaido-nippon-ham-fighters”. We’ll assume the canonical URL is https://tokyoentertainment.net/article/slug. But we don’t have exact slug. We can use placeholders but should be realistic. The gate says “replace example.com with site domain”. The current JSON-LD has “https://example.com/article” – we should change to something like “https://tokyoentertainment.net/article/softbank-vs-nipponham”. Also need to strip author if name matches placeholder. Current author is “専門編集部” – not placeholder. So keep. Also ensure FAQPage mirrors visible FAQ items only. After dedup, we have 6 items. Need to update JSON-LD to match.

    Gate 8 — Tone hygiene: Remove forbidden phrases. Scan article for any English forbidden phrases. Since it’s Japanese, likely none. But check for Japanese versions? The forbidden list is English. So skip.

    Gate 8b — Intro opener: The current intro starts with “プロ野球の対戦カードを語る時、福岡ソフトバンクホークスと北海道日本ハムファイターズの組み合わせは特別な熱を帯びます。” That is not an AI-tell opener (like “X is a”, etc.). It’s fine. Lead paragraph is 3 sentences? Actually it’s one long sentence, but it’s fine. The gate says max 2 sentences. This is one sentence. Good.

    Gate 9 — Quote speaker variety: No blockquotes, so skip.

    Gate 10 — Research confidence calibration: Research confidence is low. So rumor-list should be ≥ confirmed-list. Currently snapshot card 1 (confirmed) has 3 items, card 2 (rumor) has 3 items. After our changes from gate 3, we need to ensure balance. We’ll adjust.

    Gate 11 — Facts_summary tier audit: facts_summary is empty, so no action.

    Gate 12 — UX structural enforcement: Check requirements: comparison_table_required=true, so we need a

    with at least 3×3 matrix. The current table is already a 2-column table with many rows, but to be a 3×3 matrix, we need at least 3 columns and 3 rows. The table has only 2 columns (項目, 詳細). So it’s not a 3×3 matrix. We need to add a third column? The instruction says “≥3×3 matrix present”. The contract’s comparison_table_required is true. We need to have a table with at least 3 columns and 3 rows of comparison. Currently it’s a list. We need to reconstruct a comparison table. But we cannot change facts. We can add a third column by splitting? For example, add a column for “ソフトバンク” and “日本ハム”? That would require data. The research notes have data for both teams. We could reshape the table to compare. For instance, rows: 項目, ソフトバンク, 日本ハム. Then populate with known data: 1億円プレーヤー数, 対戦成績 etc. But that would be adding new data? Actually the article already has separate data for both teams in the body. We can extract. But we must not fabricate. The research notes have data for both teams. We can create a 3×3 comparison table. Also need at least 2 callouts (n24-tip, n24-warning). Already have tip and warning. Also need mini-summaries after H2 sections with >300 words. Check word count per H2 section. The H2 “ソフトバンクと日本ハムはどちらが勝った？” section has multiple paragraphs and a table probably >300 words. Already has n24-tldr after that section? Yes, there is a n24-tldr after that section. Good. Other sections may have >300 words? “ソフトバンクの1億円プレーヤーは何人いますか？” has short paragraphs, likely <300. "日ハムの「恐怖の9番」は誰ですか？" also short. "ソフトバンクのコーチが死亡したのはいつですか？" short. "水谷万波はハーフですか？" short. "ホークスが赤字なのはなぜですか？" longer, has a warning callout. But no n24-tldr after that section. The gate says "Mini-summary
    after any H2 section with >300 words of prose.” The last section is probably >300 words? Let’s count: it has several paragraphs and a warning. Approx 200-300? Might be borderline. But to be safe, we can add a tldr if needed. However, adding a tldr would require creating content that may not be in the article. We can write a one-line summary based on the content. But the gate says “If structures missing, reconstruct from contract + research. Do NOT fabricate facts.” We can write a summary using existing sentences. Possibly we can move the existing n24-warning? No, tldr is different. We’ll add a tldr after the last H2 section: “ホークスが赤字なのはなぜですか？”. The article ends with that section, then FAQ. So we can insert a tldr before the FAQ. The content of the tldr can be something like: “赤字の背景には高額年俸と収益のミスマッチがある。ソフトバンクはIT企業の強みを活かした効率化を進めている。” This is derived from the article.

    Also need to ensure no more than 2 consecutive

    without a break. Check current structure: there are sequences of

    tags without intervening elements. For example, after the intro, there is a stats line, then snapshot block, then table, then H2, then H3, then p, then tip, then p, etc. It’s fine.

    Gate 13 — Research-residue scan: Check for markers like “source:”, etc. None present.

    Gate 14 — Editorial voice validation: Need to verify intro first sentence takes a stance. It does take a stance: “プロ野球の対戦カードを語る時、…特別な熱を帯びます。” That’s fine. Table lead-ins: before the table there is a caption? The table is preceded by a caption in

    . But gate requires a

    with editorial framing. Currently there is no

    before the table. The table appears right after the snapshot block. We should add a

    before the table that introduces it. For example: “以下は、両チームの主要データをまとめた表です。” But that might be too bland. We can derive from the caption. Also section closers: Check each H2 section end. For example, the first H2 section (ソフトバンクと日本ハムはどちらが勝った？) ends with a n24-tldr, not a table/list/callout, so okay. Second H2 section ends with a n24-tldr as well. Third H2 section ends with n24-tip callout? Actually it ends with a paragraph and then a tip? The tip is a callout. The section ends with the tip. That violates because the last element is a callout. We need to add a closing

    after the tip. Same for other sections. Also need to check callout bodies: the existing tip bodies are not pure data recitation, they have editorial judgment. Good. Source anchor text: some anchors are like “NPB公式サイト” which is good authority. For the duplicate links we’ll adjust. Also need to ensure TL;DR editorial verdict names an actor and consequence. The existing TL;DR in first section names “ソフトバンク側は「このカードは落とさない」という自信を、日本ハム側は「差をどう埋めるか」という課題” – that names actors and consequences. Good. Need to check summary ending: the last sentence of the article before FAQ is the last paragraph of “ホークスが赤字なのはなぜですか？” section: “この強みをどう球団経営に活かすかが、今後の鍵となる。” That names actor (ソフトバンクグループ) and consequence? Vague but acceptable.

    Also need to add table lead-in p for each table. There is only one table.

    Now, implement all repairs step by step.

    First, wrap the article content in

    inside

    .

    Second, adjust snapshot cards per gate 3: move “ソフトバンクの1億円プレーヤーは7名” and “「恐怖の9番」は日本ハム・水野達稀” from card 1 to card 2 (何が不明か). Also add these to card 2 list. Need to ensure the list remains

    関連記事をさらに表示

    ジェイソン・ステイサムの経歴・代表作・最新情報を完全解説！元飛び込み選手からアクションスターへの軌跡 ラッセルクロウの現在と経歴・代表作まとめ 岡田将生の結婚・恋愛歴まとめ 鈴木アナ破局から高畑充希 【2025年最新】殺人罪の罰則と統計を徹底解説：死刑・懲役年数・強盗殺人罪との違い・死刑囚事例・年間死亡者数 【2025年最新版】Snow Man宮舘涼太（舘様）の熱愛・結婚・学歴・あだ名の真相を徹底検証する！ Xbox Game Pass 値上げ – Ultimate2,750円に 日本プラン変更まとめ 中谷美紀の経歴・プロフィールを徹底解説！渡部篠郎との関係やオーストリア移住、語学力や結婚相手、現在の生活も紹介 洞口朋子（ほらぐちともこ）の経歴・家族・当選情報・政党（中核派）・活動実績を徹底解説｜杉並区議会議員
    佐藤健一

    筆者情報

    佐藤健一

    山田太郎は東京エンターテインメントのニュース記者です。彼はエンターテインメント業界の最新情報を追い続けています。趣味は映画鑑賞と音楽制作です。

    トーキョーエントエルトアインムエント

    トーキョーエントエルトアインムエント 編集部は、出典を明確にした検証済みの報道、背景更新、訂正を公開します。

    編集デスク

    昼版 継続更新の編集サイクル

    最新記事

    最新の編集デスク更新へ素早くアクセス。

    ポリシー

    運営情報、ポリシー、読者向け連絡先。

    更新 17:48 • 昼版 • © 2026 トーキョーエントエルトアインムエント