201220231 六、發明說明 【發明所屬之技術領域】 本發明涉及資料處理技術，尤其涉及一種產品資訊的 推薦方法及系統。 【先前技術】 在網際網路技術中，網站經常需要向用戶推薦各種產 品資訊，例如電子商務網站在網頁上向用戶推薦用戶可能 感興趣的商品等。藉由這種推薦的方式，來縮短用戶尋找 所需要產品的路徑，提升用戶體驗。 一般的，網站在進行產品的推薦時，根據用戶對於某 些產品的歷史運算元據，例如用戶的產品購買歷史資料 等’使用相關性演算法確定其他產品與所購買產品之間的 關聯關係，從而將與用戶所購買的產品關聯性較強的產品 資訊推薦給用戶。 但是，這種推薦方法只考慮用戶的歷史運算元據，並 未綜合考慮其他與用戶感興趣的產品相關聯的資訊，因 此’推薦結果往往很不準確；特別地，當用戶爲新用戶 時’由於並不存在歷史運算元據，此時甚至難以爲用戶進 行產品的推薦。 而且，現有的相關性演算法本身對系統資源消耗較 大’而且，對所有的產品都需要進行與其他產品之間的關 聯關係的計算’所處理的資料量大，速度較慢，尤其是在 海量用戶、海量產品、海量訪問資料的情況下，對於資料 -5- 201220231 的處理速度緩慢，且資源消耗更爲嚴重，從而難以滿足推 薦系統的及時性要求。 【發明內容】 有鑒於此，本發明要解決的技術問題是，提供一種產 品資訊的推薦方法及系統，能夠更爲及時、準確的向用戶 推薦其可能需要的產品資訊。 爲此，本發明Η施例採用如下技術方案： 本發明實施例提供一種產品資訊的推薦方法，包括： 預先確定用戶的推薦產品集和/或產品的推薦產品 集； 獲取第一用戶的網路操作，根據第一用戶的網路操作 確定產品推薦類型； 根據確定的產品推薦類型，從第一用戶的推薦產品集 和/或所述網路操作關聯的第一產品的推薦產品集中確定 在對應的產品推薦類型下所需爲第一用戶推薦的產品資 訊。 還提供一種產品資訊的推薦系統，其特徵在於，包 括： 第一確定單元，用於預先確定用戶的推薦產品集和/ 或產品的推薦產品集； 第二確定單元，用於獲取第一用戶的網路操作，根據 第一用戶的網路操作確定產品推薦類型； 第三確定單元，用於根據確定的產品推薦類型，從第 -6- 201220231 一用戶的推薦產品集和/或所述網路操作關聯的第一產品 的推薦產品集中確定在對應的產品推薦類型下所需爲第一 用戶推薦的產品資訊。 對於上述技術方案的技術效果分析如下： 預先確定用戶和產品的推薦產品集，並且將爲用戶進 行的產品推薦分爲至少兩種推薦類型，從而根據用戶的網 路操作確定爲用戶進行推薦的產品推薦類型，進而根據產 品推薦類型確定所需爲用戶推薦的產品資訊，從而提高了 爲用戶推薦產品資訊的準確度； 而且，根據用戶的各種特性資訊、產品的特性資訊以 及用戶在一定時間段內所關注產品的資訊，據此確定每一 用戶的推薦產品集和每一產品的推薦產品集，由於在該推 薦方法中綜合考慮了用戶和產品的特性資訊，因此，推薦 結果相較於現有技術更爲合理、準確； 而且，藉由輔助推薦產品集的建立，即使新用戶進行 網路操作，或者用戶對新產品進行網路操作，也可以藉由 輔助推薦產品集基於用戶或基於產品進行產品的推薦，實 現爲新用戶或新產品進行相關產品推薦； 本發明在進行產品推薦時，僅基於預設的一個時間段 內的資料確定用戶和產品的基礎推薦產品集，而且，限定 了基礎推薦產品集的最大推薦產品數量；甚至，可以僅爲 基礎產品集數目滿足某一數目閾値的用戶，或者在一個時 間段內瀏覽次數達到某一瀏覽次數閾値的產品確定基礎推 薦產品集，從而大大減少了基礎推薦產品集的資料量，降 201220231 低了對於系統資源的要求，提高了產品推薦的速度，即使 在海量用戶、海量產品、海量產品資料的情況下，也能夠 及時地爲用戶進行產品推薦。 本發明的產品資訊推薦方法並非一定具有以上所有效 果。 【實施方式】 以下’結合附圖I羊細說明本發明產品資訊的推薦方法 及系統的0現。 在圖1所示的網路結構中，用戶藉由用戶端11與伺 服器1 2之間進行通信，以從伺服器1 2中獲取所感興趣產 品的產品資訊；並且，伺服器1 2還可以向用戶所在的用 戶端11返回向用戶推薦的產品資訊。 如圖1所示，在實際應用中，可能有多個用戶分別藉 由不同的用戶端訪問伺服器1 2。相應的，伺服器1 2需要 向每個用戶所在的用戶端返回推薦給對應用戶的產品資 訊。 如圖2所示，伺服器1 2執行以下步驟： 步驟20 1:預先確定每一用戶的推薦產品集和/或每一 產品的推薦產品集； 所述推薦產品集由若干個產品構成。所述推薦產品集 中產品數S可以自主設定，這裏並不限制。 所述推薦產品集可以包括：基礎推薦產品集和/或輔 助推薦產品集，在圖3的實施例中將詳細描述基礎推薦產 201220231 品集和輔助推薦產品集的構建方法，這裏不贅述。 步驟2〇2 :獲取第一用戶的網路操作，根據第一用戶 的網路操作確定產品推薦類型； 所述產品推薦類型可以包括：基於用戶的產品推薦和 基於產品的產品推薦。 所述基於用戶的產品推薦是指：基於用戶的偏好資訊 及歷史訪問行爲爲用戶推薦其可能感興趣的產品。 所述基於產品的產品推薦是指：基於產品的之間的相 關性’爲用戶當前關注的產品推薦相關的產品。 步驟203 :根據確定的產品推薦類型，從第一用戶的 推薦產品集和/或所述網路操作關聯的第一產品的推薦產 品集中確定在對應的產品推薦類型下所需爲第一用戶推薦 的產品資訊。 其中’當產品推薦類型爲基於用戶的產品推薦時，將 從用戶的推薦產品集中確定需要爲用戶推薦的產品資訊； 當產品推薦類型爲基於產品的產品推薦時，將從產品的推 薦產品集中確定所需爲用戶推薦的產品資訊。 圖2所示的推薦方法中，預先確定用戶和產品的推薦 產品集’並且將爲用戶進行的產品推薦分爲至少兩種推薦 類型’從而根據用戶的網路操作確定爲用戶進行推薦的產 品推薦類型’進而根據產品推薦類型確定所需爲用戶推薦 的產品資訊’從而提高了爲用戶推薦產品資訊的準確度。 以下’在圖2的基礎上藉由圖3對本發明產品推薦方 法進行更爲詳細的說明。 -9 - 201220231 如圖3所示，該方法包括： 步驟3 01 :確定每一用戶的特性資訊、每—產品的特 性資訊、每一用戶在預設的第一時間段內對產品的關注度 資訊以及每一用戶在預設的第二時間段內對產品的關注度 資訊。 每個用戶的特性資訊可以包括：用戶的來源地區，偏 好產品子類目’價格區間，品牌，風格，顏色，材質，用 戶活躍度，用戶誠信度等屬性欄位。 而每個產品的特性資訊可以包括：產品的子類目、價 格、品牌、風格、顔色、材質、資訊品質評級、熱銷度、 關注度、發佈時間等屬性欄位。 用戶對產品的關注度資訊包括：每一用戶對各種產品 的關注度値以及該用戶的來源地區。 所述第一時間段的長度可以自主設定，例如可以爲一 個月或者10天、20天等等，這裏並不限定。這裏，可以 基於用戶资訊及行爲等資料藉由統計分析和資料挖掘確定 每一用戶的特性資訊和每一產品的特性資訊。 在1Ϊ際應用中，一般可以藉由資料庫的形式分別對所 有用戶的特性資訊和所有產品的特性資訊進行儲存，例 如，建立用戶特性資料庫，以儲存每個用戶的特性資訊； 建立產品特性資料庫，以儲存每一產品的特性資訊。 步驟3 02 :根據上述資訊確定每一用戶的推薦產品集 和每一產品的推薦產品集。 具體的，每一用戶的推薦產品集可以包括：基礎推薦 -10- 201220231 產品集和/或輔助推薦產品集。 其中，每一用戶的基礎推薦產品集的確 括： 從該用戶的特性資訊中獲取用戶對應的 目；根據產品的特性資訊查找子類目屬於該 目的所有產品；從查找到的所述產品中選擇 個產品構成該用戶的基礎推薦產品集。 或者，每一用戶的基礎推薦產品集的確 括： 從該用戶的特性資訊中獲取用戶對應的 目：根據產品的特性資訊查找子類目屬於該 目的所有產品；並且， 根據各個用戶在預設的第一時間段內的 訊計算該用戶與其他用戶之間的相關性；根 預設的第二時間段內的產品關注度資訊，查 關性最高的預設第三數目個用戶在第二時間 產品； 從查找到的所有產品資訊中選擇第二預 構成該用戶的基礎推薦產品集。 其中，在確定用戶之間的相關性時，可 戶的協同過濾演算法實現。 在具體實現中，除了可以藉由預設第一 減少用戶的基礎推薦產品集確定過程中所需 外，還可以進一步對確定用戶的基礎推薦產 定方法可以包 偏好產品子類 偏好產品子類 第二預設數目 定方法可以包 偏好產品子類 偏好產品子類 產品關注度資 據各個用戶在 找與該用戶相 段內所關注的 設數目個產品 以使用基於用 時間段，以便 處理的資料量 品集這一步驟 -11 - 201220231 進行限定，從而減少用戶的基礎推薦產品集的資料量，具 體的，可以判斷所確定的用戶基礎推薦產品集中產品數目 是否超過某一預設的數目閾値，如果沒有超過，則不確定 該用戶的基礎推薦產品集，也即：對於基礎推薦產品數量 不超過某一數目閩値的用戶，不建立該用戶的基礎推薦產 品集；只有基礎推薦產品數a超過該數目閾値的用戶，才 建立該用戶的基礎推薦產品集。對於未建立基礎推薦產品 集的用戶，需要根據用戶的輔助推薦產品集進行該用戶的 產品推薦》 所述確定每一用戶的輔助推薦產品集包括： 從該用戶的特性資訊中獲取該用戶的來源地區：根據 產品的特性資訊，查找屬於該用戶的來源地區的產品中熱 銷度和/或關注度和/或發佈時間最靠前的第四預設數目個 產品構成該用戶的輔助推薦產品集。 對於每一產品，推薦產品集也可以包括：基礎推薦產 品集，或者，基礎推薦產品集和輔助推薦產品集。其中， 所述預先確定每一產品的基礎推薦結果集可以包括： 根據每一用戶在預設的第一時間段內對產品的關注度 資訊計算產品之間的相關度； 對於每一產品，選擇與該產品的相關度最高的第一預 設數目個產品構成該產品的基礎推薦產品集。 其中，在確定產品之間的相關度時可以使用產品關聯 規則推薦演算法和產品相關性推薦演算法等實現。 與用戶的基礎推薦產品集確定過程相同的’在確定產 -12- 201220231 品的基礎推薦產品集時，也可以現篩選需要建立基礎推薦 產品集的產品，具體地，可以判斷該產0·3在一預設時間段 內的瀏覽次數是否超過一預設瀏覽次數閾値，不超過時， 不爲該產品確定基礎推薦產品集；超過時，在確定該產品 的基礎推薦產品集。對於未建立基礎推薦產品集的產品， 需要藉由該產品的輔助推薦產品集確定該產品的推薦產 品。 所述確定每一用戶的輔助推薦產品集包括： 確定每一用戶的特性資訊和每一產品的特性資訊； 對於每一用戶，從該用戶的特性資訊中獲取該用戶的 來源地區；根據產品的特性資訊，查找屬於該用戶的來源 地區的產品中、熱銷度和/或關注度和/或發佈時間最靠前 的第四預設數目個產品構成該用戶的輔助推薦產品集。 所述確定產品的輔助推薦產品集包括： 根據各個用戶在預設的第一時間段內的產品關注度資 訊確定每一來源地區關注度最高的子類目下的第五預設數 目個產品構成基於產品的輔助推薦結果集。 以上的步驟3 0 1和步驟3 〇2爲伺服器爲回應用戶的網 路操作而進行的準備步驟，以下，則爲根據用戶的網路操 作而進行推薦產品的過程： 步驟3 03:獲取第一用戶的網路操作。 該第一用戶泛指任一進行網路操作的用戶。 所述網路操作可以包括：用戶打開伺服器爲用戶提供 的網頁、用戶點擊查看網頁中的某一產品、購買某—產品 -13- 201220231 等。 步驟304:根據第一用戶的網路操作確定所需爲第一 用戶提供的產品推薦類型。 其中’當用戶的網路操作不涉及產品時，則確定的產 品推薦類型一般爲：基於用戶的產品推薦，例如，用戶打 開伺服器爲用戶提供的某一網頁。 而當用戶的網路操作涉及到產品時，如用戶點擊查看 網頁中的某一產品或者購買某一產品時，則確定的產品推 薦類型可以爲：基於用戶的產品推薦和/或基於產品的產 品推薦。 當所述產品推薦類型爲基於用戶的產品推薦時，藉由 步驟3 05〜步驟306描述：當所述產品推薦類型爲基於產 品的產品推薦時，藉由步驟3 0 7〜步驟3 0 8描述。當然， 在贲際應用中將根據步驟304中所確定的產品推薦類型來 確定執行步驟3 05〜步驟306和/或步驟3 07〜步驟3 08 »並 且，當步驟304中確定兩種推薦類型都執行時，步驟 3 05〜步驟3 06和步驟3 07〜步驟3 08可以同時或者先後執 行，執行順序不限制。 步驟3 05:從第一用戶的基礎推薦產品集中獲取第六 預設數目個產品；並且，當基礎推薦產品集中產品數目小 於所述第六預設數目時，從第一用戶的輔助推薦產品集中 獲取差額個產品以獲取到所述第六預設數目個產品。 其中，當未預設輔助推薦產品集時，將不包括獲取所 述差額個產品的步驟。 -14- 201220231 步驟3 06 :將所述第六預設數目個產品按照預設第一 規則排序，選擇排序位置靠前的第七預設數目個產品作爲 所述所需爲第一用戶推薦的產品資訊。 具體的，可以根據用戶的偏好特性預設排序規則，如 符合用戶偏好的價格、品牌、風格、顔色、材質的產品優 先，並且，可以將用戶在某一段時間內已經關注過的產品 的優先順序降低，從而使得排序結果中位置靠前的產品將 更貼近用戶感興趣的產品。 步驟3 07:從第一產品的基礎推薦產品集中獲取第八 預設數目個產品：並且，當基礎推薦產品集中產品數目小 於所述第八預設數目時，從第一產品同類目的輔助推薦產 品集中獲取差額個產品以獲取到所述第八預設數目個產 品； 步驟308 :將所述第八預設數目個產品按照預設第二 規則排序，選擇排序位置靠前的第九預設數目個產品作爲 所述所需爲第一用戶推薦的產品資訊。 具體的，在進行排序時，可以根據產品之間的相關度 來進行排序，並且，可以將用戶在某一段時間內已經關注 過的產品的優先順序降低，從而使得排序結果中位置靠前 的產品將更貼近用戶感興趣的產品。 步驟309 :將所述所需爲第一用戶推薦的產品資訊向 用戶展現。 其中，由於產品推薦的類型分爲兩種，因此，在進行 推薦的產品資訊展現時，最好也根據兩種推薦類型進行區 -15- 201220231 分，以便用戶對於推薦的產品資訊更爲一目了然。 例如在電子商務網頁中，可以在用戶進入購買產品列 表時進行推薦，包括兩個產品推薦的展示欄，"購買了該 產品的用戶還購買了”展示欄展示基於產品的產品推薦類 型下得到的產品资訊，根據最後加入購買產品列表的產品 推薦與其相關的其他產品，以便實現產品之間的交叉銷 售；“其他可能感興趣的推薦”展示欄展示基於用戶的產 品推薦類型下得到的產品資訊，根據用戶的特性推薦其他 可能讓用戶感興趣的產品，進一步提升用戶的購買欲望。 另外，在實際應用中，還可以對產品的推薦效果跟蹤 評估，例如可以藉由網頁的日誌記錄獲取被推薦產品的曝 光次數，點擊次數等；或者，還可以藉由被推薦產品資料 庫的訪問交易記錄，獲取被推薦產品的回饋量，成交量。 根據下面的統計指標可評估在各交易環節推薦的準確性， 並評估推薦應用的成效，便於對推薦演算法進行優化，這 裏不赘述。 圖3所示的方法中，根據用戶的各種特性資訊、產品 的特性資訊以及用戶在一定時間段內所關注產品的資訊， 據此確定每一用戶的推薦產品集和每一產品的推薦產品 集，從而當用戶進行網路操作時，可以直接根據用戶和/ 或用戶操作的產品從用戶和/或產品對應的推薦產品集中 確定所需爲用戶推薦的產品資訊，由於在該推薦方法中綜 合考慮了用戶和產品的特性資訊，因此，推薦結果相較於 現有技術更爲準確。而且，藉由輔助推薦產品集的建立， -16- 201220231 即使新用戶進行網路操作，或者用戶對新產品進行操作， 也可以藉由輔助推薦產品集基於用戶或基於產品進行產品 的推薦，實現爲新用戶或新產品進行相關產品推薦。相對 已有的推薦系統只根據歷史操作進行推薦，本發明的推薦 結果更爲合理、準確。 另外，本發明在進行產品推薦時，僅基於預設的一個 時間段內的資料確定用戶和產品的基礎推薦產品集，而 且，限定了基礎推薦產品集的最大推薦產品數量；甚至， 可以僅爲基礎產品集數目滿足某一數目閩値的用戶，或者 在一個時間段內瀏覽次數達到某一瀏覽次數閾値的產品確 定基礎推薦產品集，從而大大減少了基礎推薦產品集的資 料量，降低了對於資源的要求，提高了產品推薦的速度， 在海量用戶、海量產品、海量產品資料的情況下，也能夠 及時地爲用戶進行產品推薦。 據統計，具有基礎推薦產品集的用戶及產品通常占到 全體用戶及產品的30%左右，進而，藉由更爲嚴格的約束 條件，如僅爲基礎產品集數目滿足某一數目閾値的用戶， 或者在一個時間段內瀏覽次數達到某一瀏覽次數閾値的產 品確定基礎推薦產品集，更是極大地縮減了用戶及產品的 基礎推薦產品集的資料量。而輔助推薦產品集是根據用戶 來源地區及產品的子類目確定的，由於用戶來源地區及產 品子類目個數一般非常有限的，因此推薦系統的性能主要 由基礎推薦產品集的資料量決定。藉由本發明的上述處 理，將基礎推薦產品集的資料量減少到全體用戶及產品量 -17- 201220231 的1 /3以下，從而大大提高了推薦系統的產品推薦速度 (可提升3-5倍，甚至更多），也解決了在海量用戶、海 fl商品、海量訪問資料的情況下產品推薦的及時性問題。 並且，藉由應用統計分析發現，在每次推薦中8 5 %以上的 用戶及產品的推薦結果來源於基礎推薦產品集，只有1 5 % 以下的新用戶、新產品的推薦結果來源於輔助推薦產品 集，因此，很好的解決了新老用戶的產品推薦問題。 與以上方法相對應的，本發明還提供一種產品資訊的 推薦系統，如圖4所示，該系統包括： 第一確定單元41，用於預先確定每一用戶的推薦產 品集和/或每一產品的推薦產品集； 第二確定單元42，用於獲取第一用戶的網路操作， 根據第一用戶的網路操作確定產品推薦類型； 第三確定單元43，用於根據確定的產品推薦類型， 從第一用戶的推薦產品集和/或所述網路操作關聯的第一 產品的推薦產品集中確定在對應的產品推薦類型下所需爲 第一用戶推薦的產品資訊。 其中，所述推薦產品集可以包括：基礎推薦產品集； 或者，所述推薦產品集包括：基礎推薦產品集和輔助推薦 產品集。 具體的，第一確定單元41可以包括： 第一確定子單元，用於確定每一用戶的推薦產品集； 和/或， 第二確定子單元，用於確定每一產品的推薦產品集。 -18- 201220231 其中，第一確定子單元可以包括·· 第一確定模組，用於確定每一用戶的特性資訊以及每 一產品的特性資訊； 第一構成模組，用於對於每一用戶，從該用戶的特性 資訊中獲取用戶對應的偏好產品子類目；根據產品的特性 =資訊查找子類目屬於該偏好產品子類目的所有產品；從査 找到的所述產品中選擇第二預設數目個產品構成該用戶的 _礎推薦產品集。 或者，第一確定子單元可以包括： 第二確定模組，用於確定每一用戶的特性資訊、每一 產品的特性資訊、用戶在預設的第一時間段內的產品關注 度資訊以及用戶在預設的第二時間段內的產品關注度資 奶； 第三確定模組，用於對於每一用戶，從該用戶的特性 資訊中獲取用戶對應的偏好產品子類目；根據產品的特性 資訊查找子類目屬於該偏好產品子類目的所有產品；並 且， 根據各個用戶在預設的第一時間段內的產品關注度資 in計算該用戶與其他用戶之間的相關性；根據各個用戶在 ii設的第二時間段內的產品關注度資訊，查找與該用戶相 關性最高的預設第三數目個用戶在第二時間段內所關注的 產品： 第二構成模組，用於從查找到的所有產品資訊中選擇 第二預設數目個產品構成該用戶的基礎推薦產品集。 -19- 201220231 第二確定子單元可以包括： 第四確定模組’用於確定每一用戶在預設的第〜時間 段內對產品的關注度資訊； 第一計算模組’用於根據所述關注度資訊計算產品之 間的相關度： 第三構成模組’用於對於每一產品，選擇與該產品的 相關度最高的第一預設數目個產品構成該產品的基礎推薦 產品集。 較佳地’第一確定子單元還可以包括： 第五確定模組’用於確定每一用戶的特性資訊和每一 產品的特性資訊； 第四構成模組，用於對於每一用戶，從該用戶的特性 資訊中獲取該用戶的來源地區；根據產品的特性資訊，查 找屬於該用戶的來源地區的產品中、熱銷度和/或關注度 和/或發佈時間最筇前的第四預設數目個產品構成該用戶 的輔助推薦產品集。 較佳地，第二確定子單元還可以包括： 第五構成模組，用於根據各個用戶在預設的第一時間 段內的產品關注度資訊確定每一來源地區關注度最高的子 類目下的第五預設數目個產品構成基於產品的輔助推薦結 果集。 其中，所述產品推薦類型包括：基於用戶的產品推薦 和基於產品的產品推薦’此時’ 當所述產品推薦類型爲基於用戶的產品推薦時’第三 -20- 201220231 確定單元43可以包括： 第一獲取子單元，用於從第一用戶的基礎推薦產品集 中獲取第六預設數目個產品；並且，當基礎推薦產品集中 產品數目小於所述第六預設數目時，從第一用戶的輔助推 薦產品集中獲取差額個產品以獲取到所述第六預設數目個 產品； 第一選擇子單元，用於將所述第六預設數目個產品按 照預設第一規則排序，選擇排序位置靠前的第七預設數目 個產品作爲所述所需爲第一用戶推薦的產品資訊。 或者，當所述產品推薦類型爲基於產品的產品推薦 時，第三確定單元43可以包括： 第二獲取子單元，用於從第一產品的基礎推薦產品集 中獲取第八預設數目個產品；並且，當基礎推薦產品集中 產品數目小於所述第八預設數目時，從第一產品同類目的 輔助推薦產品集中獲取差額個產品以獲取到所述第八預設 數目個產品； 第二選擇子單元，用於將所述第八預設數目個產品按 照預設第二規則排序，選擇排序位置靠前的第九預設數目 個產品作爲所述所需爲第一用戶推薦的產品資訊。 較佳地，該系統還可以包括： 展現單元44，用於將所述所需爲第一用戶推薦的產 品資訊向用戶展現。 對於以上的產品推薦系統，第一確定單元預先確定用 戶和產品的推薦產品集，並且將爲用戶進行的產品推薦分 -21 - 201220231 爲至少兩種推薦類型，從而第二確定單元根據用戶的網路 操作確定爲用戶進行推薦的產品推薦類型，進而第三確定 單元根據產品推薦類型確定所需爲用戶推薦的產品資訊， 從而提高了爲用戶推薦產品資訊的準確度； 而且，根據用戶的各種特性資訊、產品的特性資訊以 及用戶在一定時間段內所關注產品的資訊，據此確定每一 用戶的推薦產品集和每一產品的推薦產品集，由於在該推 薦系統中綜合考慮了用戶和產品的特性資訊，因此，推薦 結果相較於現有技術更爲合理、準確； 而且，藉由輔助推薦產品集的建立，即使新用戶進行 網路操作，或者用戶對新產品進行網路操作，也可以藉由 輔助推薦產品集基於用戶或基於產品進行產品的推薦，實 現爲新用戶或新產品進行相關產品推薦》 在以上的本發明贸施例中，包括第一預設數目、第二 預設數目...第八預設數目等多個預設的資料，這些資料之 間並沒有必然的聯繫，在實際應用中，各個資料的數値可 以相同也可以不同，這裏並不限定。 本領域普通技術人員可以理解，實現上述實施例的方 法的過程可以藉由程式指令相關的硬體來完成，所述的程 式可以儲存於可讀取儲存媒體中，該程式在執行時執行上 述方法中的對應步驟。所述的儲存媒體可以如： ROM/RAM、磁碟、光碟等。 以上所述僅是本發明的較佳實施方式，應當指出，對 於本技術領域的普通技術人員來說，在不脫離本發明原理 -22- 201220231 的前提下’還可以做出若干改進和潤飾，這些改進和潤飾 也應視爲本發明的保護範圍。 【圖式簡單說明】 圖1爲本發明應用場景下的網路結構示例； 圖2爲本發明一種產品資訊的推薦方法流程示意圖： 圖3爲本發明另一種產品資訊的推薦方法流程示意 圖， 圖4爲本發明一種產品資訊的推薦系統結構示意圖。 【主要元件符號說明】 1 1 :用戶端 1 2 :伺服器 41 :第一確定單元 42 :第二確定單元 43 :第三確定單元 44 :展現單元 -23-201220231 VI. Description of the Invention [Technical Field] The present invention relates to data processing technology, and in particular, to a method and system for recommending product information. [Prior Art] In Internet technology, websites often need to recommend various product information to users. For example, an e-commerce website recommends products that may be of interest to users on a web page. With this recommended method, the user's path to find the desired product is shortened, and the user experience is improved. Generally, when the website performs the recommendation of the product, the user uses the historical algorithm for certain products, such as the user's product purchase history data, to determine the relationship between other products and the purchased product. Therefore, the product information with strong relevance to the product purchased by the user is recommended to the user. However, this recommendation method only considers the user's historical operation metadata, and does not comprehensively consider other information associated with the product that the user is interested in. Therefore, the recommendation result is often very inaccurate; in particular, when the user is a new user' Since there is no historical operation metadata, it is even difficult to recommend the product to the user at this time. Moreover, the existing correlation algorithm itself consumes a large amount of system resources, and the calculation of the relationship between all products and other products requires a large amount of data, which is slow, especially in In the case of massive users, massive products, and massive access data, the processing speed of the data-5-201220231 is slow and the resource consumption is more serious, which makes it difficult to meet the timeliness requirements of the recommended system. SUMMARY OF THE INVENTION In view of the above, the technical problem to be solved by the present invention is to provide a method and system for recommending product information, which can recommend the product information that may be needed to the user in a timely and accurate manner. To this end, the present invention adopts the following technical solutions: The embodiment of the present invention provides a method for recommending product information, including: predetermining a recommended product set of a user and/or a recommended product set of the product; acquiring a network of the first user An operation, determining a product recommendation type according to the network operation of the first user; determining, according to the determined product recommendation type, the recommended product set of the first user and/or the recommended product concentration of the first product associated with the network operation The product information recommended by the first user is required under the product recommendation type. A recommendation system for providing product information, comprising: a first determining unit, configured to predetermine a recommended product set of the user and/or a recommended product set of the product; and a second determining unit, configured to acquire the first user The network operation determines the product recommendation type according to the network operation of the first user; the third determining unit is configured to select a product set of the user from the -6-201220231 and/or the network according to the determined product recommendation type The recommended product set of the first product associated with the operation determines the product information recommended for the first user under the corresponding product recommendation type. The technical effects of the above technical solutions are analyzed as follows: The recommended product set of the user and the product is determined in advance, and the product recommendation for the user is divided into at least two recommended types, thereby determining the recommended product for the user according to the user's network operation. The recommended type, and then the product information recommended by the user is determined according to the product recommendation type, thereby improving the accuracy of recommending the product information for the user; and, according to the user's various characteristic information, product characteristic information, and the user within a certain period of time According to the information of the products concerned, the recommended product set of each user and the recommended product set of each product are determined accordingly, and since the user and the product characteristic information are comprehensively considered in the recommended method, the recommendation result is compared with the prior art. More reasonable and accurate; Moreover, by establishing the recommended product set, even if the new user performs network operations, or the user performs network operations on the new product, the product can be based on the user or the product based on the auxiliary recommended product set. Recommendations for new users or new products Related product recommendation; The present invention determines the basic recommended product set of the user and the product based on the data within a preset time period when the product is recommended, and defines the maximum recommended product quantity of the basic recommended product set; Only the number of users whose basic product set number meets a certain threshold threshold, or the product recommendation basic product set whose number of browsing times reaches a certain threshold of browsing times in a certain period of time, thereby greatly reducing the amount of data of the basic recommended product set, and reducing 201220231 The requirements for system resources are lowered, and the speed of product recommendation is improved. Even in the case of massive users, mass products, and massive product data, product recommendations can be made for users in a timely manner. The product information recommendation method of the present invention does not necessarily have the above-mentioned effective effects. [Embodiment] The following describes the recommended method and system of the product information of the present invention in conjunction with FIG. In the network structure shown in FIG. 1, the user communicates with the server 12 by the client 11 to obtain product information of the product of interest from the server 12; and the server 12 can also The product information recommended to the user is returned to the user terminal 11 where the user is located. As shown in FIG. 1, in an actual application, there may be multiple users accessing the server 12 by different clients. Correspondingly, the server 12 needs to return the product information recommended to the corresponding user to the user end where each user is located. As shown in FIG. 2, the server 12 performs the following steps: Step 20: Predetermine a recommended product set for each user and/or a recommended product set for each product; the recommended product set is composed of several products. The number of products S in the recommended product set can be set autonomously, and is not limited herein. The recommended product set may include: a basic recommended product set and/or a supplementary recommended product set. The construction method of the basic recommended product 201220231 and the auxiliary recommended product set will be described in detail in the embodiment of FIG. 3, and details are not described herein. Step 2: 2: Obtain a network operation of the first user, and determine a product recommendation type according to a network operation of the first user; the product recommendation type may include: a product recommendation based on the user and a product recommendation based on the product. The user-based product recommendation refers to recommending a product that may be of interest to the user based on the user's preference information and historical access behavior. The product-based product recommendation refers to: based on the correlation between products, the relevant products are recommended for the products currently of interest to the user. Step 203: Determine, according to the determined product recommendation type, the recommended product set of the first user and/or the recommended product set of the first product associated with the network operation, and determine the first user recommendation under the corresponding product recommendation type. Product information. Where 'when the product recommendation type is based on the user's product recommendation, the product information that needs to be recommended for the user will be determined from the user's recommended product set; when the product recommendation type is based on the product-based product recommendation, the recommended product set of the product will be determined Product information recommended for the user. In the recommendation method shown in FIG. 2, the recommended product set of the user and the product is determined in advance and the product recommendation for the user is divided into at least two types of recommendation' to determine the recommended product recommendation for the user according to the user's network operation. The type 'and then determines the product information that is recommended for the user based on the type of product recommendation' thus improves the accuracy of recommending product information to the user. The following is a more detailed description of the product recommendation method of the present invention based on Fig. 2 on the basis of Fig. 2. -9 - 201220231 As shown in FIG. 3, the method includes: Step 3: Determine characteristic information of each user, characteristic information of each product, and attention of each user to the product in a preset first time period. Information and information about each user's attention to the product during the preset second time period. The characteristics information of each user may include: the user's source area, preferred product sub-category 'price range, brand, style, color, material, user activity, user integrity and other attribute fields. The product information of each product can include: sub-category, price, brand, style, color, material, information quality rating, hot sales, attention, release time and other attribute fields of the product. User's attention to the product includes: each user's attention to various products and the source area of the user. The length of the first period of time may be set autonomously, for example, one month or 10 days, 20 days, etc., and is not limited herein. Here, it is possible to determine the characteristic information of each user and the characteristic information of each product by statistical analysis and data mining based on information such as user information and behavior. In a one-time application, the characteristic information of all users and the characteristic information of all products can be stored in the form of a database, for example, a user characteristic database is established to store characteristic information of each user; A database to store information about the characteristics of each product. Step 3 02: Determine the recommended product set of each user and the recommended product set of each product based on the above information. Specifically, each user's recommended product set may include: basic recommendation -10- 201220231 product set and/or auxiliary recommended product set. Wherein, each user's basic recommended product set is determined to: obtain the user's corresponding content from the user's characteristic information; and find the subcategory belongs to all products of the purpose according to the product characteristic information; select from the found products The products constitute the basic recommended product set for the user. Or, each user's basic recommended product set is determined to: obtain the user's corresponding content from the user's characteristic information: according to the product characteristic information, find the sub-category belongs to all products of the purpose; and, according to each user, the preset The information in the first time period calculates the correlation between the user and other users; the product attention information in the second time period preset by the root, and the preset third number of users with the highest visibility in the second time Product; select the second pre-constituted product set of the user from all the product information found. Among them, when determining the correlation between users, the user's collaborative filtering algorithm is implemented. In a specific implementation, in addition to determining the first recommended product set by the first reduction user, the basic recommendation production method for determining the user may further include a product sub-category preference product sub-category. The second preset number determining method may include a preference product sub-category product sub-category product attention degree data, each user is looking for a set number of products that are concerned with the user phase segment to use the time period based on the time period for processing The product step -11 - 201220231 is limited to reduce the amount of data of the user's basic recommended product set. Specifically, it can be determined whether the determined number of products in the user-based recommended product product exceeds a certain preset number threshold. If there is no more than, the user's basic recommended product set is not determined, that is, for the user whose basic recommended product quantity does not exceed a certain number, the basic recommended product set of the user is not established; only the basic recommended product number a exceeds the A user with a threshold number of users is required to establish a basic recommended product set for the user. For a user who does not have a basic recommended product set, the user recommended product selection according to the user's auxiliary recommended product set is determined. The user recommended product set for each user includes: obtaining the user's source from the user's characteristic information. Region: According to the product characteristic information, find the fourth preset number of products in the products belonging to the source region of the user, and/or the top predetermined number of products that constitute the user's auxiliary recommended product set. . For each product, the recommended product set may also include: a base recommended product set, or a base recommended product set and an auxiliary recommended product set. The pre-determining the basic recommendation result set of each product may include: calculating, according to each user's attention information of the product in the preset first time period, the correlation between the products; for each product, selecting The first predetermined number of products that are most relevant to the product constitute the basic recommended product set for the product. Among them, the product association rule recommendation algorithm and the product correlation recommendation algorithm can be implemented when determining the correlation between products. The same as the user's basic recommended product set determination process, when determining the basic recommended product set of the production-12-201220231 product, it is also possible to screen the products that need to establish the basic recommended product set. Specifically, the product can be judged to be 0. Whether the number of views in a preset time period exceeds a preset number of browsing thresholds, and when not exceeded, the basic recommended product set is not determined for the product; when exceeded, the basic recommended product set of the product is determined. For products that do not have a basic recommended product set, the recommended product for that product needs to be determined by the product's auxiliary recommended product set. The determining the auxiliary recommended product set of each user includes: determining characteristic information of each user and characteristic information of each product; for each user, obtaining a source area of the user from the characteristic information of the user; The feature information finds that the fourth predetermined number of products belonging to the source region of the user, the popularity and/or the degree of attention and/or the release time constitutes the user's auxiliary recommended product set. The determining the recommended product set of the product comprises: determining, according to the product attention information of each user in the preset first time period, the fifth preset number of products under the sub-category with the highest degree of interest in each source region is based on Auxiliary recommendation result set for the product. The above steps 3 0 1 and 3 〇 2 are preparation steps for the server to respond to the user's network operation. The following is the process of recommending the product according to the user's network operation: Step 3 03: Get the first A user's network operation. The first user refers to any user who performs network operations. The network operation may include: the user opens a webpage provided by the server for the user, the user clicks to view a certain product in the webpage, purchases a certain product-13-201220231, and the like. Step 304: Determine a product recommendation type that is required to be provided for the first user according to the network operation of the first user. Where 'when the user's network operation does not involve the product, the determined product recommendation type is generally: based on the user's product recommendation, for example, the user opens a certain webpage provided by the server for the user. When the user's network operation involves the product, such as when the user clicks to view a certain product in the webpage or purchases a certain product, the determined product recommendation type may be: based on the user's product recommendation and/or product-based product. recommend. When the product recommendation type is a user-based product recommendation, it is described by steps 305 to 306: when the product recommendation type is a product-based product recommendation, it is described by steps 3 0 7 to 3 0 8 . Of course, in the inter-application, it will be determined according to the product recommendation type determined in step 304 to perform step 3 05 to step 306 and/or step 3 07 to step 3 08 » and, when step 304, it is determined that both types of recommendation are When executed, steps 3 05 to 3 06 and steps 3 07 to 3 08 can be executed simultaneously or sequentially, and the execution order is not limited. Step 3 05: Obtain a sixth preset number of products from the first recommended product set of the first user; and, when the number of products in the basic recommended product set is less than the sixth preset number, select the recommended product from the first user Obtaining a difference of products to obtain the sixth predetermined number of products. Wherein, when the auxiliary recommended product set is not preset, the step of obtaining the difference product will not be included. -14- 201220231 Step 3 06: Sort the sixth preset number of products according to a preset first rule, and select a seventh preset number of products with the top position as the recommended one for the first user. Product Information. Specifically, the ranking rule may be preset according to the preference characteristics of the user, such as the product priority of the price, brand, style, color, and material that meets the user's preference, and the priority of the product that the user has paid attention to in a certain period of time may be prioritized. Reduced so that the top-ranked products in the sorting results will be closer to the product of interest to the user. Step 3: Obtain an eighth preset number of products from the basic recommended product set of the first product: and, when the number of products in the basic recommended product set is less than the eighth preset number, the auxiliary recommended product from the first product of the same type Collecting the difference products to obtain the eighth preset number of products; Step 308: Sorting the eighth preset number of products according to the preset second rule, and selecting the ninth preset number of the top position The products are the product information recommended for the first user. Specifically, when sorting, the sorting may be performed according to the correlation between the products, and the priority order of the products that the user has paid attention to in a certain period of time may be lowered, so that the products with the top positions in the sorting result are obtained. Will be closer to the product of interest to the user. Step 309: Present the product information recommended for the first user to the user. Among them, since the types of product recommendation are divided into two types, it is better to perform the district -15-201220231 points according to the two recommended types when recommending the product information presentation, so that the user can see the recommended product information more clearly. For example, in an e-commerce webpage, a recommendation can be made when a user enters a list of purchased products, including a display bar for two product recommendations, and a user who has purchased the product also purchases a display bar to display a product-based product recommendation type. Product information, based on the products that are last added to the list of purchased products, and other related products, in order to achieve cross-selling between products; "Other recommendations that may be of interest" display bar displays products based on the user's product recommendation type Information, according to the characteristics of the user, recommend other products that may be of interest to the user, further enhance the user's desire to purchase. In addition, in the actual application, the recommendation effect of the product can also be tracked and evaluated, for example, the log record of the webpage can be obtained. The recommended number of exposures, clicks, etc.; or, by accessing the transaction records of the recommended product database, the feedback amount and volume of the recommended products can be obtained. According to the following statistical indicators, the recommended in each transaction link can be evaluated. Accuracy, and evaluate the recommended application The effect is easy to optimize the recommendation algorithm, which is not described here. In the method shown in Figure 3, according to the user's various characteristic information, product characteristic information and the information of the product of the user in a certain period of time, it is determined accordingly. A recommended product set of a user and a recommended product set of each product, so that when the user performs network operation, the required product set can be determined directly from the user and/or the product corresponding to the product according to the user and/or the product operated by the user. The product information recommended by the user, because the user and the product characteristic information are comprehensively considered in the recommended method, the recommendation result is more accurate than the prior art. Moreover, by the establishment of the auxiliary recommended product set, -16- 201220231 Even if a new user performs network operations, or if the user operates on a new product, the product recommendation based on the user or based on the product can be recommended by the auxiliary recommended product set to implement related product recommendation for the new user or the new product. The recommendation system only recommends based on historical operations, and the recommended results of the present invention are more reasonable. In addition, when performing product recommendation, the present invention determines the basic recommended product set of the user and the product based only on the data within a preset time period, and defines the maximum recommended product quantity of the basic recommended product set; Only the number of basic product sets meets a certain number of users, or the product recommendation base product set that reaches the threshold of a certain number of browsing times within a certain period of time, thereby greatly reducing the amount of data of the basic recommended product set and reducing The requirements for resources have increased the speed of product recommendation. In the case of massive users, mass products, and massive product data, it is also possible to recommend products to users in a timely manner. According to statistics, users and products with basic recommended product sets are usually Accounted for about 30% of all users and products, and then, with more stringent constraints, such as only a certain number of basic product sets to meet a certain number of threshold users, or a number of views in a period of time to reach a certain number of views The threshold product is determined by the basic recommended product set, which is greatly reduced. The recommended amount of data base product set of users and products. The auxiliary recommended product set is determined according to the user source area and the subcategory of the product. Since the number of user source areas and product subcategories is generally very limited, the performance of the recommended system is mainly determined by the amount of data of the basic recommended product set. . By the above processing of the present invention, the amount of data of the basic recommended product set is reduced to less than 1/3 of the total user and product quantity -17-201220231, thereby greatly improving the product recommendation speed of the recommendation system (up to 3-5 times, Even more, it also solves the problem of timeliness of product recommendation in the case of massive users, sea fl products, and massive access data. Moreover, by applying statistical analysis, it is found that the recommendation results of more than 85% of users and products in each recommendation are from the basic recommended product set, and only the recommended results of new users and new products below 15% are from the auxiliary recommendation. The product set, therefore, is a good solution to the product recommendation problem of new and old users. Corresponding to the above method, the present invention further provides a product information recommendation system. As shown in FIG. 4, the system includes: a first determining unit 41, configured to predetermine each user's recommended product set and/or each The recommended product set of the product; the second determining unit 42 is configured to obtain the network operation of the first user, determine the product recommendation type according to the network operation of the first user, and the third determining unit 43 is configured to determine the type of the product according to the determined product Determining product information recommended for the first user under the corresponding product recommendation type from the recommended product set of the first user and/or the recommended product set of the first product associated with the network operation. The recommended product set may include: a basic recommended product set; or the recommended product set includes: a basic recommended product set and an auxiliary recommended product set. Specifically, the first determining unit 41 may include: a first determining subunit, configured to determine a recommended product set of each user; and/or a second determining subunit, configured to determine a recommended product set of each product. -18- 201220231 The first determining subunit may include: a first determining module, configured to determine characteristic information of each user and characteristic information of each product; and a first constituent module, configured for each user Obtaining a sub-category of the preferred product corresponding to the user from the characteristic information of the user; searching for all products belonging to the sub-category of the preferred product according to the characteristics of the product=information; selecting the second pre-selected product from the found products Set a number of products to form the user's _ basic recommended product set. Alternatively, the first determining subunit may include: a second determining module, configured to determine characteristic information of each user, characteristic information of each product, product attention information of the user in a preset first time period, and a user The product attention degree milk in the preset second time period; the third determining module is configured to obtain, for each user, the preferred product subcategory corresponding to the user from the characteristic information of the user; according to the characteristics of the product The information search subcategory belongs to all products of the sub-category of the preferred product; and, according to the product attention degree of each user in the preset first time period, the correlation between the user and other users is calculated; according to each user The product attention information in the second time period set by ii is used to find the product of the third time that the user has the highest relevance to the user in the second time period: the second component module is used to The second predetermined number of products selected among all the product information found constitutes the basic recommended product set of the user. -19- 201220231 The second determining subunit may include: a fourth determining module 'for determining information about the degree of attention of each user in the preset first time period; the first calculating module' is used according to the The degree of relevance between the attention information calculation products: The third component module 'for each product, selects the first predetermined number of products with the highest degree of relevance to the product to constitute a basic recommended product set of the product. Preferably, the first determining subunit may further include: a fifth determining module configured to determine characteristic information of each user and characteristic information of each product; and a fourth constituent module, configured for each user The user's characteristic information obtains the source area of the user; according to the product characteristic information, finds the product in the source area of the user, the popularity and/or attention and/or the fourth pre-release time A set of products constitutes a set of auxiliary recommended products for the user. Preferably, the second determining sub-unit may further include: a fifth component module, configured to determine, according to the product attention information of each user in the preset first time period, the sub-category with the highest degree of attention in each source region The fifth predetermined number of products constitutes a product-based auxiliary recommendation result set. The product recommendation type includes: a user-based product recommendation and a product-based product recommendation 'at this time'. When the product recommendation type is a user-based product recommendation, the third -20-201220231 determining unit 43 may include: a first obtaining subunit, configured to obtain a sixth preset number of products from a basic recommended product set of the first user; and, when the number of products in the basic recommended product set is less than the sixth preset number, from the first user The auxiliary recommendation product collects the difference products to obtain the sixth preset number of products; the first selection sub-unit is configured to sort the sixth preset number of products according to the preset first rule, and select the sorting position. The seventh predetermined number of products in the front are used as the product information recommended by the first user. Alternatively, when the product recommendation type is a product-based product recommendation, the third determining unit 43 may include: a second obtaining sub-unit, configured to obtain an eighth preset number of products from the basic recommended product set of the first product; And, when the number of products in the basic recommended product set is less than the eighth preset number, the difference products are obtained from the first product auxiliary recommended products of the first product to obtain the eighth preset number of products; And a unit, configured to sort the eighth preset number of products according to a preset second rule, and select a ninth preset number of products with a top position as the product information recommended by the first user. Preferably, the system may further include: a presentation unit 44, configured to present the product information recommended by the first user to the user. For the above product recommendation system, the first determining unit predetermines the recommended product set of the user and the product, and the product recommendation for the user is divided into 21 - 201220231 as at least two recommended types, so that the second determining unit is based on the user's network. The road operation determines the recommended product recommendation type for the user, and the third determining unit determines the product information recommended by the user according to the product recommendation type, thereby improving the accuracy of recommending the product information for the user; and, according to various characteristics of the user Information, product characteristics information, and information about the products that users pay attention to within a certain period of time, based on which each user's recommended product set and recommended product set for each product are determined, because users and products are comprehensively considered in the recommendation system. The characteristic information, therefore, the recommendation result is more reasonable and accurate than the prior art; Moreover, by establishing the auxiliary recommended product set, even if the new user performs network operation, or the user performs network operation on the new product, By assisting the recommended product set based on the user or based on the product Recommendation of the product, to implement the relevant product recommendation for the new user or new product. In the above-mentioned trade example of the present invention, the first preset number, the second preset number, the eighth preset number, and the like are included. There is no necessary connection between the materials and the data. In practical applications, the number of data may be the same or different, and is not limited here. A person skilled in the art can understand that the process of implementing the method of the foregoing embodiment can be completed by using a program instruction related hardware, and the program can be stored in a readable storage medium, and the program executes the above method when executed. The corresponding steps in . The storage medium may be, for example, a ROM/RAM, a magnetic disk, a compact disk, or the like. The above description is only a preferred embodiment of the present invention, and it should be noted that those skilled in the art can make several improvements and refinements without departing from the principle of the present invention -22-201220231. These improvements and finishes should also be considered as protection of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a schematic diagram of a network structure in an application scenario according to the present invention; FIG. 2 is a schematic flowchart of a method for recommending product information according to the present invention: FIG. 3 is a schematic flowchart of another method for recommending product information according to the present invention. 4 is a schematic structural diagram of a recommendation system of product information of the present invention. [Description of main component symbols] 1 1 : User terminal 1 2 : Server 41 : First determining unit 42 : Second determining unit 43 : Third determining unit 44 : Presentation unit -23-