A new study revealed that men and women tend to behave differently when it comes to contacting potential partners in dating websites.

The study, published in the journal Social Network Analysis and Mining, suggests that men tend to be more aggressive and contact other users they are interested in while women tend to be more conscious of their own attractiveness to other users.

For the study, the researchers developed a reciprocal recommendation system that better matches users who are mutually interested in and likely to communicate with each other. They based their new system on data collected from Baihe, one of the largest dating websites in China. The new reciprocal recommendation system calculates a reciprocal score that measures the compatibility between two users and each potential dating candidate, generating a recommendation list to include users with top scores.

The researchers then tested their new method in Baihe and discovered that their proposed recommendation system significantly outperformed previously proposed approaches. Additionally, the collaborative filtering-based algorithms achieve much better performance than content-based algorithms in both precision and recall.

Their study also showed behavioral differences between male and female users when it comes to sending a message to their potential matches. Male users tend to focused on their own interest and are oblivious to their attractiveness to potential dates while female users tend to be more conscious toward their attractiveness and take it into consideration when finding potential dates.

"We found that males like to send a lot of messages to attractive female users, but they don't get a lot of responses," explained Shuangfei Zhai, a PhD candidate at Binghamton University and co-author of the study, in a statement. "For females, they're self-conscious because they tend to evaluate the likelihood of getting a response to the user that they're sending messages to. In terms of the data, it shows that women have a much larger chance of getting responses from users that they send messages to."