企業間あるいは業種間の取引関係や産業間の投入・産出構造など，オブジェクト間あるいはカテゴリ間の相互関係を重み付きグラフと見なして効果的に可視化することは，全体構造や法則性を把握するのに有用な方法の1つと考えられている．一方，重み付きグラフの可視化法として球面可視化法（SE-PI-W法）があるが，重みの規模の格差が大きい場合，規模の大きな重みが支配的となり，解釈可能な可視化結果を得ることが困難な場合がある．本稿では，オブジェクト間のインタラクションを数え上げカテゴリに関するMixing Matrixを構築し，インタラクション数が多項分布に従うとし，ランダム性を仮定したモデルによる期待値および標準偏差によりZスコアを計算することで，カテゴリ間の依存度合の重みとしてZスコアを用いた可視化法を提案する．企業間取引ネットワークおよび人工データを用いた実験より，規模の大きさにより隠れてしまう取引関係の強さを表現した可視化結果が得られることを示す．さらに，重みの変化などのノイズに対する可視化結果の頑健性について，定量的に評価する．

For aiming at uncovering whole structures or rules of mutual connections, it is considered to be one of the useful approaches to visualize effectively relationships between objects or categories such as business relations between enterprises or industries. We treat these multiple or weighted relationships as weighted graphs. On the other hand, it may be difficult to obtain the interpretable visualization results by an existing weighted graphs visualization method (SE-PI-W method) when disparities of scales of multiplicities or weights are large because large-scale weights may become dominant. In this paper, we propose an effective visualization method for these weighted graphs. Our method first builts the Mixing Matrix from the number of interactions between categories, second calculates Z scores by expected values and standard deviations under assuming that the values of the Mixing Matrix distributed according to a multinomial distribution. Our method treats categories as nodes and uses the Z scores as weights between categories, and visualizes these weighted graphs by the SE-PI-W method. From our experimental evaluations using a inter-firm networks and artificial data, we show that the visualization results which can uncover strengths and weaknesses of business connections of hiding with the size of scales. Furthermore, we evalute robustness of our method with respect to noise such as changes of degree of weight added on each link.