Java点餐小程序实现个性化推荐菜品

编程入门行业动态更新时间:2024-10-22 11:05:07

Java点餐小程序实现个性化推荐<a href=https://www.elefans.com/category/jswz/34/1717585.html style= 菜品"/>

Java点餐小程序实现个性化推荐菜品

1、业务数据表说明

顾客与订单是一对多的关系；订单与菜品是多对多的关系，所以还有一张关系表menu_order_；

2、从数据库查询并构建训练集

现在需要从这四张表联合查询查出每一个顾客的下单菜品的数量。

1）存着每个顾客的下单菜品的菜品id和此菜品的下单次数。

2）先查出所有顾客customers，再查出每一个顾客的下单菜品的菜品id和此菜品的下单次数，写出到文件中（filename1是不带评分的数据，即userId-dishId；而filename2是userId-dishId-value）。

@PostConstruct
public void getUserItems() {try (Writer w1 = new BufferedWriter(new FileWriter(filename1));Writer w2 = new BufferedWriter(new FileWriter(filename2))) {// 查询所有用户List<Customer> customers = customerMapper.selectList(null);// 查询所有用户的下单菜品记录for (Customer customer : customers) {List<DishNumber> dishNumbers = orderMapper.getUserOrderDishes(customer.getCId());if (dishNumbers.size() == 0)continue;for (DishNumber d : dishNumbers) {w1.write(customer.getId() + "," + d.getDId() + "\n");w2.write(customer.getId() + "," + d.getDId() + "," + d.getValue() + "\n");}}w1.flush();w2.flush();} catch (IOException e) {e.printStackTrace();}
}

getUserDishes的sql语句：根据cId查询结果。

<select id="getUserOrderDishes" resultType="com.firstGroup.restaurant.model.vo.DishNumber">select d.id dId, count(*) valuefrom menu_order_ mo, order_ o, customer c, dish dwhere o.c_id=c.c_id and mo.o_id=o.o_id and d.d_id=mo.d_id and o.c_id = #{cId}group by d.d_id
</select>

3）生成的训练集（数据时随意测试的，没有参考性）

数据说明：顾客id=1：下单菜品id=1，61次；菜品id=2，1次；菜品id=3，1次

filename1； filename2

3、使用Mahout实现基于用户的协调过滤推荐算法

1）有用户评分的：

/** @Author Haojie* @Description 有评分的基于用户的协同过滤推荐算法* @Param* @return**/
private List<Dish> userCFWithoutRecommend(String cId, Integer number) throws Exception {// 建立数据模型，包含用户评分DataModel dm = new GenericDataModel(GenericDataModel.toDataMap(new FileDataModel(new File(filename2))));// 使用曼哈顿距离计算类似度UserSimilarity us = new EuclideanDistanceSimilarity(dm);//指定NearestNUserNeighborhood做为近邻算法UserNeighborhood unb = new NearestNUserNeighborhood(10, us, dm);// 构建包含用户评分的UserCF推荐器Recommender re = new GenericUserBasedRecommender(dm, unb, us);// 返回推荐结果，为cId用户推荐number个商品Integer id = customerMapper.selectById(cId).getId();List<RecommendedItem> list = re.recommend(id, number);List<Dish> dishes = new ArrayList<>();System.out.println("根据用户cId=" + cId + ", userId=" + id + "的点餐习惯，推荐的前" + number + "个菜品的id和推荐度如下：");for (RecommendedItem recommendedItem : list) {System.out.println(recommendedItem.getItemID() + " : " + recommendedItem.getValue());dishes.add(dishMapper.selectOne(new LambdaQueryWrapper<Dish>().eq(Dish::getId,recommendedItem.getItemID())));}return dishes;
}

2）无用户评分的：

/** @Author Haojie* @Description 无评分的基于用户的协同过滤推荐算法* @Param* @return**/
private List<Dish> userCFWithoutScoreRecommend(String cId, Integer number) throws Exception {// 建立数据模型，不包含用户评分DataModel dm = new GenericDataModel(GenericDataModel.toDataMap(new FileDataModel(new File(filename1))));// 使用曼哈顿距离计算类似度UserSimilarity us = new CityBlockSimilarity(dm);//指定NearestNUserNeighborhood做为近邻算法UserNeighborhood unb = new NearestNUserNeighborhood(5, us, dm);// 构建不包含用户评分的UserCF推荐器Recommender re = new GenericBooleanPrefUserBasedRecommender(dm, unb, us);// 返回推荐结果，为cId用户推荐number个商品Integer id = customerMapper.selectById(cId).getId();List<RecommendedItem> list = re.recommend(id, number);List<Dish> dishes = new ArrayList<>();System.out.println("根据用户cId=" + cId + ", userId=" + id + "的点餐习惯，推荐的前" + number + "个菜品的id和推荐度如下：");for (RecommendedItem recommendedItem : list) {System.out.println(recommendedItem.getItemID() + " : " + recommendedItem.getValue());dishes.add(dishMapper.selectOne(new LambdaQueryWrapper<Dish>().eq(Dish::getId,recommendedItem.getItemID())));}return dishes;
}

最终都可以返回推荐的菜品数据。