Music suggested by members

15 posts / 0 new

Sun, 21/02/2021 - 11:50

Editor in search of Anningan & Malina

Joined: 10.05.2012

Pending moderation

Hello everybody!
I've been thinking about this for some time now. LT is based mainly on two things: music and languages. My impression is that users tend to use the website more for its linguistical advantages (it's clear how LT can be a valid language learning tool), whereas guests are here more for the music. I'm obviously generalising, but my idea is to make this aspect more prominent for users as well.
There are different ways we can discover new music. The "Trending Now" and "New Content" sections are really useful in this regard, especially because they're shown on the homepage. There are more though: we can check the latest translations, our friends feed, look for songs in a particular language or genre and so on. We have many tools to discover new music, but I think we could take advantage of another one: suggested music.
The idea is to create a special section in the home page for music suggested by members. Each member would have the chance to put their favourite song under the spotlight for others to listen to it. I was thinking of a very simple form where each members puts the link to the song they want everyone to listen to*. Maybe we could do this every first day of the month, so the list changes every month and people don't abuse the system by suggesting a million songs just because they're obsessed with them. It would be also useful for this section to appear both on the homepage as well as on the side when browsing through the website (like the "Site Activity" box).

I've got some reasons on why this could be a big advantage:
1) It will bring more visibility to the website. Guests will come to LT not only to look for a translation of a song they like, but also to discover new music they may enjoy;
2) It may convince more guests to become translators (I'm talking in particular for those languages like Chinese or Japanese that are really popular, although their presence on LT is minimal) or even just to use LT more often. This will increase the user engagement and boost LT's reputation;
3) As a consequence, this will boost the website revenues. I know, profit shouldn't come first, but without ads, the website wouldn't even exist;
4) It will help translators finding new songs to translate (I acknowledge not everyone has this problem, but I do);
5) It will introduce people to the music of different countries. This happens already, but because there are some languages that are more popular than others, there's no much variety. On the "Trending Now" section I often see American, Russian, European and Korean songs. It would be nice to be introduced to "unpopular" music;
6) It will make translators use the site more often and interact with each other more. Everyone is a little self-centred when it comes to introduce other people to the things they like and enjoy. Different people might discover they have the same music taste and build friendships. We've all met friends on LT, this will be just an additional way to meet other people.

I want to say something very important now. What I like about LT is that it's a little space where I can feel safe. I often come here to distract myself from real life problems. Sometimes I spend hours browsing through the website, even if I don't publish anything (as I said, I often struggle to find something to translate). The admins are very humble people who see this website as a hobby. This view is shared by us viewers and I think this is what makes LT special: there's no pressure, no desire to make something remarkable or outstanding. It's just a niche little space lost in the vastness of the Internet and I like it for this exact reason. LT doesn't try to be anything more than what it already is. I can see some people thinking that my suggestion may be unnecessary. We are not on yt, LT is not a company that is trying to get as much profit as possible, to get a reputation in its field and become a point of reference for song lyrics.
I had this idea because I love this website so much (and I'm sure you feel the same) and I would like it to grow and see it get more recognition. I don't want LT to grow to the point of becoming a giant corporation, but I see that the users are always the same and, although we all appreciate their contribution and help, I think getting more people would be a big advantage to make the website more complete. As I said, some sections of this website are clearly more popular than others and I think it would be nicer to change things for the better and promote more music than we already do.

What do you think about this?

*it should be obvious, but I'll make myself clear just to be on the safe side: the song should already be present on the website. Every link should redirect to other pages on LT.

Sun, 21/02/2021 - 12:33

hey_doc

Super Member

Joined: 23.04.2016

How would this be any different from creating a thread for music, or randomly clicking in site activity, or the trending now box?

You could just let people post songs in a thread.
User 1 posts song in thread
User 2 has to listen to the first song, comment on it, and then post their own.
User 3 does the same.
Repeat.

I don't understand the dynamics of submitting a song to the homepage. How many songs get displayed and how are they chosen? How big is the list?
I think song submission is not interactive at all and not at all different from finding a new song on your own.

Sun, 21/02/2021 - 12:55

Alma Barroca

Moderator ᐂ

Joined: 05.04.2012

That would be something to think of - I think we need to expand interaction among users here, bring a personal touch to it. Of course, for those that are interested in it. We have the chatrooms, but not many people use (I myself often do not). But I really liked your suggestion, Joshua

Wed, 24/02/2021 - 18:40

Flopsi

Editor Soldier of Love

Joined: 12.03.2017

Hi Joshua,

is this all about finding "new" music to translate? Really? I've got so much songs to translate I don't even know where to start. Reminds me of my children... "I'm bored. Is there anything I could do?" "Yeah, sure, the windows need cleaning, the trash needs to be taken out. You could clean the bathroom, vacuum the living room..." "Nah, I wanted something fun to do."

I check the translation requests, the charts, the last comments. I listen to the radio. I translate everything. Songs I love, translations I've seen that could need some improvement, songs that make me curious and of course all songs of The Jacksons and The Cure.

But, yeah, I like your suggestion. It would be nice to make people aware of songs or lyrics that aren't so popular and that you loved to translate.

Wed, 24/02/2021 - 21:31

Schnurrbrat

Russian asset

Joined: 07.03.2019

Hi, Joshua.
I'm not sure that I'm on the same page with you, just as Juan said your suggestion needs to be processed.

I once suggested something related: to add "points of popularity" to songs, just like thanks now. But these will count towards the popularity of the song among members. Currently, popularity is affected by recent fluctuations of interest, so some relatively unknown songs have a chance to take the top spots and to distract a new user/guest who is unfamiliar with this artist.

Back to your suggestion: highlighting forgotten songs (or songs that you like) sounds like a good idea to me, but I'm afraid we will succumb to K-pop wave then (based on topnodes). I frankly don't know how it will work out. Let say I will highlight some Soviet song from the 60s, but no one will listen to it, I'm certain of this. With so many languages, genres and styles it will be difficult to please a random guest.

Besides, this website has so many features (sometimes hidden in plain sight) that it could be difficult for a random guest to process everything.

Thu, 25/02/2021 - 02:22

OgingerO

Senior Member

Joined: 08.09.2017

One feature from your list of suggestions that do think will might be useful, is "other bands liked by fans of this band" - it could be implemented in different ways, for example, if many people "heart" a band as a favourite AND also heart another band, maybe there could be a little thing, see other bands that are liked by fans of this band. (For an example of this in action, although it's based on songs rather than bands, is https://www.last.fm/

I think liking bands rather than songs is the way to go, because a lot of people might like a certain popular song by an artist, but not the artist themselves - and if you do it by songs you'll get recommendations for these songs/bands you'd never enjoy in a million years (say a band whose song was featured on a tv show that a lot of people liked.. but those people might just like that song in the show... and generally listen to another genre of music... you wind up getting awful recommendations this way. Youtube music uses this it seems, in it's algorithm as well (it's awful).. so if a punk band does.. a parody of say... Celine Dion, and a lot of Dion fans like it because it's a cover of her music... Punk fans are going to get recommendations for Celine Dion... yeah, not pretty. But bands that are liked by the same fans.. it's not perfect, but it's a better way to go.

Thu, 25/02/2021 - 03:12

(Reply to #6)

Schnurrbrat

Russian asset

Joined: 07.03.2019

OgingerO,
That would be nice indeed to make this mechanics automated, but I'm afraid that algorithms behind YT or Pinterest suggestions are not that simple to implement.

Thu, 25/02/2021 - 04:43

(Reply to #7)

OgingerO

Senior Member

Joined: 08.09.2017

Sorry, what I meant was, if there was a way you could see, as a list, what other users favourited -
like if a bunch of people favourited band A. On band A's page next to the "Add to favorites" heart, there was a link to "see what bands other users who liked this band also liked" and then it could just show a list of the most common bands that other users, who also liked Band A also liked. So if you liked Band A, B, C, F, Z and I liked Band A, C, H, G and I, and a third guy liked Band A, B, M, W and I, and a fourth guy liked: it would show, users liked (other than the initial Band A) Users also liked: B and B, C, I - (AKA only the bands multiple other users liked). It wouldn't be much at the beginning, as it doesn't seem like (IMHO) that favoriting bands is a big thing (Like Bob Dylan has 209 songs translated 822 times to 47 languages, but only 17 favourited him., BTS (Bangtan Boys) (236 songs translated 3372 times to 61 languages) only 217 favourited them) - but over time it would become a familiar and welcome thing as more people would want to see what other bands are suggested by this feature . As far as I can tell, you can click on the list to see who likes a band/artist, but once you go to their profile you can't see who else they liked. (Even if you could it would be a clunky way to see other bands you might like). So it wouldn't be overly complex to be able just to add, cross referencing of band favourites to have say the most liked other bands by people who favourited the same band.

Thoughts?

Thu, 25/02/2021 - 05:55

(Reply to #8)

Schnurrbrat

Russian asset

Joined: 07.03.2019

Art this point I can only repeat myself by saying it would be nice to have, but I have to clue how difficult it is to implement. And I believe that Joshua's suggestion was more of users' ability to place songs they like in the limelight. I'm sure that Admins and tech wizards have some thoughts about this too. Let's keep our fingers crossed and wait for new developments. On the other hand, one of things that attracted me to this website is my discovery of some absolutely amazing, yet rarely known lyrics, poems and artists. I would like to see more of those in my feed rather than mainstream titles.

#10

Thu, 25/02/2021 - 06:26

OgingerO

Senior Member

Joined: 08.09.2017

I don't know how difficult it would be either - but I can't imagine it would be overly difficult. I'm not really a fan of users recommendations in the vein of spotlighting songs randomly. Unless it's in, if you like this - you might like this. I mean, I don't care if someone randomly suggests a song - I don't know about their other tastes so how can I know if I think they have terrible or awesome taste in music?

Since we can't see people's favourites - there's no way to tell. I don't know about you, but it seems like he is suggesting something similar to when the bookstore has an endcap that says suggestions from our employees.. .I think it might work better (The suggestion) if it was genre based, but since LT's genre options are ridiculous (lacking even the most basic categorization of genres while including ones that most people would know if offered a million dollars) it wouldn't be very useful. I mean there's Austropop but not Britpop. Éntekhno but not Chamber, Orchestral or Baroque Pop, Nasheed gets it's own category, but Gospel and Jewish, etc., don't. (That's a whole other issue with LT however, ).

#11

Thu, 25/02/2021 - 08:24

BlueBird

Super Member

Joined: 27.12.2020

Seems we are talking about something like "you may also like" or "see also" (suggested similar content) and popularity (e.g. popularity of a movie in IMDb). 'what are the best songs/movies/books/etc of 2020/January/etc" or "top ten/20/50/100/.." things. Some of the first things many people usually search for when being introduced into a field.

One sticky issue seems to be the mechanism of compiling such lists. Based on the movies experience, the two main choices are by direct voting (IMDb) or by algorithmic aggregation (e.g. weighted sum) of the critiques ratings (most notably rottentomatoes and metacritic). Some hybrid ones (e.g. CinemaScore, surveying audience by a professional organization, probably doing some statistical processing, removing outliers, etc) are also possible.

IMDb rating is basically a raw measure of popularity. Rotten tomatoes and metacritic scores are more a measure of how 'artistically/critically' good the movie is. Many times these two are not consistent. Google results, Wikipedia, etc seems to give more weight to 'rottentomatoes' and 'metacritic'.

The first method is fairly easy (just set up a poll, use thanks, votes etc) but prone to uninformed or biased ratings. The second one is a bit more difficult(defining and finding the 'critiques' in LT, giving an objective fair 'weight' to each one, and choosing the aggregation algorithm) but if implemented well, can be a more reliable rating.

Personally I prefer the second method, but I understand it might not be easy to implement. Once the rating issue is handled, expansion of the notion (categories, etc) can add more value to the final product.

#12

Thu, 25/02/2021 - 08:29

(Reply to #11)

OgingerO

Senior Member

Joined: 08.09.2017

I'm talking more along the lines of https://www.gnooks.com/ if you know it, or https://www.literature-map.com/ (same guy), goodreads does it too, but not as good.

#13

Thu, 25/02/2021 - 09:34

(Reply to #12)

BlueBird

Super Member

Joined: 27.12.2020

Very interesting sites. Thank you for letting me know about that.

I understand it is taking it a step further. Instead of recommending content (book, music, movie, etc) for a general audience, it is about finding what content is best for a particular person, i.e. content personalization. Another interesting project here.

It is a very exciting and at the same time challenging idea. We might be talking about some special branch of Artificial Intelligence(AI), i.e. expert systems. I also understand AI is used extensively in online business and marketing, what is usually called Business Intelligence. Basically it is gathering and analyzing data based on the user interaction with an online site. For ex, part of it, in an online store check what pages, items the user clicks on, how long he/she remains in a page, etc and then analyze that data (usually by some sort of data scientist) to find the behavior patterns for that user and suggest content that the user might be interested on. Another famous one is personalized ads by Google.

In LT, it would mean - within the limits of privacy- analyzing the interaction of the user with the site. For ex. having a history of the pages he had visited, how long he has stayed in each page, what particular singer, band, time period, etc his choices are, what is the statistical distribution of his thanks, votes, comments, etc. Based on all this, predicting some behavior (I won't dare say personality) model of the user and based on that suggest contents to him (in the form of a list, a map, etc).

Bringing AI to this is very exciting and at the same time challenging. As you know, General AI, having a machine that think and feel exactly like a human might be more of a dream for the time being, although it may come true some time in the future. Expert systems do good job in more analytical jobs, but to my understanding not that good when it comes to cognition and particularly, to emotions. And implementing them are not trivial. Can be a real technological challenge.

The advantage is personalization of the content, the disadvantage I think is first, the person might sort of get caught in his own web (just following the same pattern). In human-based systems (esp. critique-based ones), we have the pleasant opportunity to get an informed idea about the new content coming up. Don't know much about the sites you kindly introduced me to yet (just did some basic jobs), but I guess they are probably aware of the issue and might even already have some fixes to mitigate the issue (or maybe even solve it to some extent).

Another issue is how good is AI in mimicking the 'informed common sense' rating we get by the algorithmic aggregation of the critiques ratings. It is a matter of how good is the Expert system (AI system trying to translate a human expert's knowledge into some rules and then use it to do some deduction and getting to some results, here rating a content and finding how suitable it is to a particular user's taste).A sticky issue here might be feelings and emotions, since art is very dependent on emotions and AI in general is known to be not as good at that as in more concrete and analytical thinking.

Thanks again for very thought-provoking ideas and discussion. And also for reading this far

#14

Thu, 25/02/2021 - 21:30

Anerneq

Editor in search of Anningan & Malina

Joined: 10.05.2012

hey_doc wrote:

How would this be any different from creating a thread for music, or randomly clicking in site activity, or the trending now box?

I don't understand the dynamics of submitting a song to the homepage. How many songs get displayed and how are they chosen? How big is the list?

As I said, there are many ways to do something similar to this. My idea was to make it part of the system, instead of having to do it by yourself by filtering the songs or creating a thread for the purpose. Threads are good, but after a while they get lost. New threads get created and old ones easily go wither away.
My idea was to make a little form available to every user. It would be one song per user, so theoretically everyone could contribute, but not everyone will be interested. So yeah, there's no limit to the length of the list. This could be a problem, because it defeats the purpose of putting songs you like under the spotlight for the others, but mine is just a suggestion and it can be improved (or disregarded).

Flopsi wrote:

is this all about finding "new" music to translate? Really?

The main idea is to promote your musical tastes. I had this idea because I always see the same kind of songs in the Popular section and because sometimes I may translate a song that I consider worth listening, but my translations don't get enough views, so I assume the songs don't get listened either. It could actually be used as a way to promote your translations, but, as I wrote, there are many more advantages and reasons to this.

Schnurrbrat wrote:

Back to your suggestion: highlighting forgotten songs (or songs that you like) sounds like a good idea to me, but I'm afraid we will succumb to K-pop wave then (based on topnodes). I frankly don't know how it will work out. Let say I will highlight some Soviet song from the 60s, but no one will listen to it, I'm certain of this. With so many languages, genres and styles it will be difficult to please a random guest.

Besides, this website has so many features (sometimes hidden in plain sight) that it could be difficult for a random guest to process everything.

That's the reason why I thought of promoting these songs on the homepage. I understand guests don't browse the site filtering the songs to find something they're interested in.
I can't deny that we may risk being flooded by Kpop, but our community is very diverse: the idea was to have a much broader variety of music thanks to how diverse our community is. It's true that an old Soviet song might go unnoticed among Kpop hits, but the opposite is also true: your contribution might be noticed by some exactly, because it's something unusual. Maybe putting a little "note" to explain why others should listen to the song you suggested might make things better? It would be a way to attract people more easily, but it may also be too much.

#15

Sat, 03/04/2021 - 09:44

Flopsi

Editor Soldier of Love

Joined: 12.03.2017

If the main idea is to promote musical tastes it's simple - I love R'n'B, but also all kinds of music as long as it's good, like this one - nice music & nice lyrics:

https://lyricstranslate.com/en/i-am-moon-ich-bin-der-mond.html

Someone else who'd like to share?